Mercurial > repos > galaxyp > mqppep_preproc

annotate mqppep_mrgfltr.py @ 1:b76c75521d91 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
author galaxyp
date Fri, 28 Oct 2022 18:26:42 +0000
parents 8dfd5d2b5903
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1 #!/usr/bin/env python
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
2
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
3 # Import the packages needed
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
4 import argparse
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
5 import operator # for operator.itemgetter
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
6 import os.path
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
7 import re
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
8 import shutil # for shutil.copyfile(src, dest)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
9 import sqlite3 as sql
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
10 import sys # import the sys module for exc_info
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
11 import time
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
12 import traceback # for formatting stack-trace
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
13 from codecs import getreader as cx_getreader
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
14
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
15 import numpy as np
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
16 import pandas
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
17
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
18 # global constants
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
19 N_A = "N/A"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
20
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
21
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
22 # ref: https://stackoverflow.com/a/8915613/15509512
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
23 # answers: "How to handle exceptions in a list comprehensions"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
24 # usage:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
25 # from math import log
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
26 # eggs = [1,3,0,3,2]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
27 # print([x for x in [catch(log, egg) for egg in eggs] if x is not None])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
28 # producing:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
29 # for <built-in function log>
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
30 # with args (0,)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
31 # exception: math domain error
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
32 # [0.0, 1.0986122886681098, 1.0986122886681098, 0.6931471805599453]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
33 def catch(func, *args, handle=lambda e: e, **kwargs):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
34
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
35 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
36 return func(*args, **kwargs)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
37 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
38 print("For %s" % str(func))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
39 print(" with args %s" % str(args))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
40 print(" caught exception: %s" % str(e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
41 (ty, va, tb) = sys.exc_info()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
42 print(" stack trace: " + str(traceback.format_exception(ty, va, tb)))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
43 exit(-1)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
44 return None
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
45
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
46
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
47 def whine(func, *args, handle=lambda e: e, **kwargs):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
48
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
49 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
50 return func(*args, **kwargs)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
51 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
52 print("Warning: For %s" % str(func))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
53 print(" with args %s" % str(args))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
54 print(" caught exception: %s" % str(e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
55 (ty, va, tb) = sys.exc_info()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
56 print(" stack trace: " + str(traceback.format_exception(ty, va, tb)))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
57 return None
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
58
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
59
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
60 def ppep_join(x):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
61 x = [i for i in x if N_A != i]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
62 result = "%s" % " | ".join(x)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
63 if result != "":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
64 return result
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
65 else:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
66 return N_A
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
67
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
68
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
69 def melt_join(x):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
70 tmp = {key.lower(): key for key in x}
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
71 result = "%s" % " | ".join([tmp[key] for key in tmp])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
72 return result
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
73
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
74
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
75 def __main__():
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
76 # Parse Command Line
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
77 parser = argparse.ArgumentParser(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
78 description="Phopsphoproteomic Enrichment Pipeline Merge and Filter."
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
79 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
80
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
81 # inputs:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
82 # Phosphopeptide data for experimental results, including the intensities
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
83 # and the mapping to kinase domains, in tabular format.
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
84 parser.add_argument(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
85 "--phosphopeptides",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
86 "-p",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
87 nargs=1,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
88 required=True,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
89 dest="phosphopeptides",
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
90 help=" ".join([
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
91 "Phosphopeptide data for experimental results, including the",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
92 "intensities and the mapping to kinase domains, in tabular format"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
93 ]),
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
94 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
95 # UniProtKB/SwissProt DB input, SQLite
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
96 parser.add_argument(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
97 "--ppep_mapping_db",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
98 "-d",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
99 nargs=1,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
100 required=True,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
101 dest="ppep_mapping_db",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
102 help="UniProtKB/SwissProt SQLite Database",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
103 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
104 # species to limit records chosed from PhosPhositesPlus
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
105 parser.add_argument(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
106 "--species",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
107 "-x",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
108 nargs=1,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
109 required=False,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
110 default=[],
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
111 dest="species",
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
112 help=" ".join([
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
113 "limit PhosphoSitePlus records to indicated species",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
114 "(field may be empty)"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
115 ]),
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
116 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
117
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
118 # outputs:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
119 # tabular output
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
120 parser.add_argument(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
121 "--mrgfltr_tab",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
122 "-o",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
123 nargs=1,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
124 required=True,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
125 dest="mrgfltr_tab",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
126 help="Tabular output file for results",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
127 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
128 # CSV output
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
129 parser.add_argument(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
130 "--mrgfltr_csv",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
131 "-c",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
132 nargs=1,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
133 required=True,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
134 dest="mrgfltr_csv",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
135 help="CSV output file for results",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
136 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
137 # SQLite output
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
138 parser.add_argument(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
139 "--mrgfltr_sqlite",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
140 "-S",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
141 nargs=1,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
142 required=True,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
143 dest="mrgfltr_sqlite",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
144 help="SQLite output file for results",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
145 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
146
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
147 # "Make it so!" (parse the arguments)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
148 options = parser.parse_args()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
149 print("options: " + str(options))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
150
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
151 # determine phosphopeptide ("upstream map") input tabular file access
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
152 if options.phosphopeptides is None:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
153 exit('Argument "phosphopeptides" is required but not supplied')
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
154 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
155 upstream_map_filename_tab = os.path.abspath(options.phosphopeptides[0])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
156 input_file = open(upstream_map_filename_tab, "r")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
157 input_file.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
158 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
159 exit("Error parsing phosphopeptides argument: %s" % str(e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
160
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
161 # determine input SQLite access
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
162 if options.ppep_mapping_db is None:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
163 exit('Argument "ppep_mapping_db" is required but not supplied')
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
164 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
165 uniprot_sqlite = os.path.abspath(options.ppep_mapping_db[0])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
166 input_file = open(uniprot_sqlite, "rb")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
167 input_file.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
168 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
169 exit("Error parsing ppep_mapping_db argument: %s" % str(e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
170
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
171 # copy input SQLite dataset to output SQLite dataset
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
172 if options.mrgfltr_sqlite is None:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
173 exit('Argument "mrgfltr_sqlite" is required but not supplied')
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
174 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
175 output_sqlite = os.path.abspath(options.mrgfltr_sqlite[0])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
176 shutil.copyfile(uniprot_sqlite, output_sqlite)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
177 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
178 exit("Error copying ppep_mapping_db to mrgfltr_sqlite: %s" % str(e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
179
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
180 # determine species to limit records from PSP_Regulatory_Sites
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
181 if options.species is None:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
182 exit(
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
183 'Argument "species" is required (& may be empty) but not supplied'
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
184 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
185 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
186 if len(options.species) > 0:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
187 species = options.species[0]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
188 else:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
189 species = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
190 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
191 exit("Error parsing species argument: %s" % str(e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
192
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
193 # determine tabular output destination
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
194 if options.mrgfltr_tab is None:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
195 exit('Argument "mrgfltr_tab" is required but not supplied')
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
196 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
197 output_filename_tab = os.path.abspath(options.mrgfltr_tab[0])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
198 output_file = open(output_filename_tab, "w")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
199 output_file.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
200 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
201 exit("Error parsing mrgfltr_tab argument: %s" % str(e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
202
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
203 # determine CSV output destination
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
204 if options.mrgfltr_csv is None:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
205 exit('Argument "mrgfltr_csv" is required but not supplied')
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
206 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
207 output_filename_csv = os.path.abspath(options.mrgfltr_csv[0])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
208 output_file = open(output_filename_csv, "w")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
209 output_file.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
210 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
211 exit("Error parsing mrgfltr_csv argument: %s" % str(e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
212
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
213 def mqpep_getswissprot():
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
214
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
215 #
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
216 # copied from Excel Output Script.ipynb BEGIN #
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
217 #
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
218
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
219 # String Constants #################
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
220 DEPHOSPHOPEP = "DephosphoPep"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
221 DESCRIPTION = "Description"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
222 FUNCTION_PHOSPHORESIDUE = (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
223 "Function Phosphoresidue(PSP=PhosphoSitePlus.org)"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
224 )
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
225 # Gene Name from UniProtKB
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
226 GENE_NAME = "Gene_Name"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
227 # ON_FUNCTION column from PSP_Regulatory_Sites
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
228 ON_FUNCTION = ("ON_FUNCTION")
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
229 # NOTES column from PSP_Regulatory_Sites
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
230 ON_NOTES = "NOTES"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
231 # ON_OTHER_INTERACT column from PSP_Regulatory_Sites
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
232 ON_OTHER_INTERACT = "ON_OTHER_INTERACT"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
233 # ON_PROCESS column from PSP_Regulatory_Sites
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
234 ON_PROCESS = ("ON_PROCESS")
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
235 # ON_PROT_INTERACT column from PSP_Regulatory_Sites
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
236 ON_PROT_INTERACT = "ON_PROT_INTERACT"
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
237 PHOSPHOPEPTIDE = "Phosphopeptide"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
238 PHOSPHOPEPTIDE_MATCH = "Phosphopeptide_match"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
239 PHOSPHORESIDUE = "Phosphoresidue"
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
240 PUTATIVE_UPSTREAM_DOMAINS = " ".join([
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
241 "Putative Upstream Kinases(PSP=PhosphoSitePlus.org)/",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
242 "Phosphatases/Binding Domains"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
243 ])
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
244 SEQUENCE = "Sequence"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
245 SEQUENCE10 = "Sequence10"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
246 SEQUENCE7 = "Sequence7"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
247 SITE_PLUSMINUS_7AA_SQL = "SITE_PLUSMINUS_7AA"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
248 UNIPROT_ID = "UniProt_ID"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
249 UNIPROT_SEQ_AND_META_SQL = """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
250 select Uniprot_ID, Description, Gene_Name, Sequence,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
251 Organism_Name, Organism_ID, PE, SV
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
252 from UniProtKB
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
253 order by Sequence, UniProt_ID
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
254 """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
255 UNIPROT_UNIQUE_SEQ_SQL = """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
256 select distinct Sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
257 from UniProtKB
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
258 group by Sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
259 """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
260 PPEP_PEP_UNIPROTSEQ_SQL = """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
261 select distinct phosphopeptide, peptide, sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
262 from uniprotkb_pep_ppep_view
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
263 order by sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
264 """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
265 PPEP_MELT_SQL = """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
266 SELECT DISTINCT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
267 phospho_peptide AS 'p_peptide',
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
268 kinase_map AS 'characterization',
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
269 'X' AS 'X'
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
270 FROM ppep_gene_site_view
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
271 """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
272 # CREATE TABLE PSP_Regulatory_site (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
273 # site_plusminus_7AA TEXT PRIMARY KEY ON CONFLICT IGNORE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
274 # domain TEXT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
275 # ON_FUNCTION TEXT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
276 # ON_PROCESS TEXT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
277 # ON_PROT_INTERACT TEXT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
278 # ON_OTHER_INTERACT TEXT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
279 # notes TEXT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
280 # organism TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
281 # );
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
282 PSP_REGSITE_SQL = """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
283 SELECT DISTINCT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
284 SITE_PLUSMINUS_7AA ,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
285 DOMAIN ,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
286 ON_FUNCTION ,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
287 ON_PROCESS ,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
288 ON_PROT_INTERACT ,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
289 ON_OTHER_INTERACT ,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
290 NOTES ,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
291 ORGANISM
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
292 FROM PSP_Regulatory_site
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
293 """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
294 PPEP_ID_SQL = """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
295 SELECT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
296 id AS 'ppep_id',
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
297 seq AS 'ppep_seq'
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
298 FROM ppep
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
299 """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
300 MRGFLTR_DDL = """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
301 DROP VIEW IF EXISTS mrgfltr_metadata_view;
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
302 DROP TABLE IF EXISTS mrgfltr_metadata;
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
303 CREATE TABLE mrgfltr_metadata
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
304 ( ppep_id INTEGER REFERENCES ppep(id)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
305 , Sequence10 TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
306 , Sequence7 TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
307 , GeneName TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
308 , Phosphoresidue TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
309 , UniProtID TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
310 , Description TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
311 , FunctionPhosphoresidue TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
312 , PutativeUpstreamDomains TEXT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
313 , PRIMARY KEY (ppep_id) ON CONFLICT IGNORE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
314 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
315 ;
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
316 CREATE VIEW mrgfltr_metadata_view AS
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
317 SELECT DISTINCT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
318 ppep.seq AS phospho_peptide
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
319 , Sequence10
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
320 , Sequence7
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
321 , GeneName
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
322 , Phosphoresidue
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
323 , UniProtID
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
324 , Description
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
325 , FunctionPhosphoresidue
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
326 , PutativeUpstreamDomains
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
327 FROM
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
328 ppep, mrgfltr_metadata
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
329 WHERE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
330 mrgfltr_metadata.ppep_id = ppep.id
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
331 ORDER BY
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
332 ppep.seq
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
333 ;
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
334 """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
335
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
336 CITATION_INSERT_STMT = """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
337 INSERT INTO Citation (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
338 ObjectName,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
339 CitationData
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
340 ) VALUES (?,?)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
341 """
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
342 CITATION_INSERT_PSP = " ".join([
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
343 "PhosphoSitePlus(R) (PSP) was created by Cell Signaling",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
344 "Technology Inc. It is licensed under a Creative Commons",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
345 "Attribution-NonCommercial-ShareAlike 3.0 Unported License.",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
346 "When using PSP data or analyses in printed publications or",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
347 "in online resources, the following acknowledgements must be",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
348 "included: (a) the words",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
349 '"PhosphoSitePlus(R), www.phosphosite.org" must',
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
350 "be included at appropriate places in the text or webpage,",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
351 "and (b) the following citation must be included in the",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
352 'bibliography: "Hornbeck PV, Zhang B, Murray B, Kornhauser',
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
353 "JM, Latham V, Skrzypek E PhosphoSitePlus, 2014: mutations,",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
354 "PTMs and recalibrations. Nucleic Acids Res. 2015",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
355 '43:D512-20. PMID: 25514926."'
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
356 ])
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
357 CITATION_INSERT_PSP_REF = " ".join([
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
358 'Hornbeck, 2014, "PhosphoSitePlus, 2014: mutations, PTMs and',
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
359 'recalibrations.", https://pubmed.ncbi.nlm.nih.gov/22135298,',
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
360 "https://doi.org/10.1093/nar/gkr1122"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
361 ])
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
362
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
363 MRGFLTR_METADATA_COLUMNS = [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
364 "ppep_id",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
365 "Sequence10",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
366 "Sequence7",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
367 "GeneName",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
368 "Phosphoresidue",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
369 "UniProtID",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
370 "Description",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
371 "FunctionPhosphoresidue",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
372 "PutativeUpstreamDomains",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
373 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
374
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
375 # String Constants (end) ############
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
376
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
377 class Error(Exception):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
378 """Base class for exceptions in this module."""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
379
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
380 pass
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
381
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
382 class PreconditionError(Error):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
383 """Exception raised for errors in the input.
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
384
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
385 Attributes:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
386 expression -- input expression in which the error occurred
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
387 message -- explanation of the error
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
388 """
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
389
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
390 def __init__(self, expression, message):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
391 self.expression = expression
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
392 self.message = message
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
393
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
394 # start_time = time.clock() #timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
395 start_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
396
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
397 # get keys from upstream tabular file using readline()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
398 # ref: https://stackoverflow.com/a/16713581/15509512
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
399 # answer to "Use codecs to read file with correct encoding"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
400 file1_encoded = open(upstream_map_filename_tab, "rb")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
401 file1 = cx_getreader("latin-1")(file1_encoded)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
402
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
403 count = 0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
404 upstream_map_p_peptide_list = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
405 re_tab = re.compile("^[^\t]*")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
406 while True:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
407 count += 1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
408 # Get next line from file
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
409 line = file1.readline()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
410 # if line is empty
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
411 # end of file is reached
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
412 if not line:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
413 break
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
414 if count > 1:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
415 m = re_tab.match(line)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
416 upstream_map_p_peptide_list.append(m[0])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
417 file1.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
418 file1_encoded.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
419
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
420 # Get the list of phosphopeptides with the p's that represent
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
421 # the phosphorylation sites removed
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
422 re_phos = re.compile("p")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
423
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
424 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
425 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
426 "%0.6f pre-read-SwissProt [0.1]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
427 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
428 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
429
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
430 # -------- Get SwissProt data from SQLite database (start) -----------
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
431 # build UniProt sequence LUT and list of unique SwissProt sequences
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
432
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
433 # Open SwissProt SQLite database
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
434 conn = sql.connect(uniprot_sqlite)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
435 cur = conn.cursor()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
436
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
437 # Set up structures to hold SwissProt data
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
438
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
439 uniprot_Sequence_List = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
440 UniProtSeqLUT = {}
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
441
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
442 # Execute query for unique seqs without fetching the results yet
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
443 uniprot_unique_seq_cur = cur.execute(UNIPROT_UNIQUE_SEQ_SQL)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
444
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
445 while 1:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
446 batch = uniprot_unique_seq_cur.fetchmany(size=50)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
447 if not batch:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
448 # handle case where no records are returned
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
449 break
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
450 for row in batch:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
451 Sequence = row[0]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
452 UniProtSeqLUT[(Sequence, DESCRIPTION)] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
453 UniProtSeqLUT[(Sequence, GENE_NAME)] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
454 UniProtSeqLUT[(Sequence, UNIPROT_ID)] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
455 UniProtSeqLUT[Sequence] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
456
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
457 # Execute query for seqs and metadata without fetching the results yet
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
458 uniprot_seq_and_meta = cur.execute(UNIPROT_SEQ_AND_META_SQL)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
459
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
460 while 1:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
461 batch = uniprot_seq_and_meta.fetchmany(size=50)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
462 if not batch:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
463 # handle case where no records are returned
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
464 break
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
465 for (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
466 UniProt_ID,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
467 Description,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
468 Gene_Name,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
469 Sequence,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
470 OS,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
471 OX,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
472 PE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
473 SV,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
474 ) in batch:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
475 uniprot_Sequence_List.append(Sequence)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
476 UniProtSeqLUT[Sequence] = Sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
477 UniProtSeqLUT[(Sequence, UNIPROT_ID)].append(UniProt_ID)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
478 UniProtSeqLUT[(Sequence, GENE_NAME)].append(Gene_Name)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
479 if OS != N_A:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
480 Description += " OS=" + OS
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
481 if OX != -1:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
482 Description += " OX=" + str(OX)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
483 if Gene_Name != N_A:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
484 Description += " GN=" + Gene_Name
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
485 if PE != N_A:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
486 Description += " PE=" + PE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
487 if SV != N_A:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
488 Description += " SV=" + SV
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
489 UniProtSeqLUT[(Sequence, DESCRIPTION)].append(Description)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
490
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
491 # Close SwissProt SQLite database; clean up local variables
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
492 conn.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
493 Sequence = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
494 UniProt_ID = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
495 Description = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
496 Gene_Name = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
497
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
498 # -------- Get SwissProt data from SQLite database (finish) -----------
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
499
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
500 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
501 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
502 "%0.6f post-read-SwissProt [0.2]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
503 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
504 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
505
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
506 # -------- Get SwissProt data from SQLite database (start) -----------
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
507 # Open SwissProt SQLite database
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
508 conn = sql.connect(uniprot_sqlite)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
509 cur = conn.cursor()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
510
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
511 # Set up dictionary to aggregate results for phosphopeptides
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
512 # corresponding to dephosphoeptide
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
513 DephosphoPep_UniProtSeq_LUT = {}
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
514
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
515 # Set up dictionary to accumulate results
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
516 PhosphoPep_UniProtSeq_LUT = {}
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
517
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
518 # Execute query for tuples without fetching the results yet
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
519 ppep_pep_uniprotseq_cur = cur.execute(PPEP_PEP_UNIPROTSEQ_SQL)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
520
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
521 while 1:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
522 batch = ppep_pep_uniprotseq_cur.fetchmany(size=50)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
523 if not batch:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
524 # handle case where no records are returned
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
525 break
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
526 for (phospho_pep, dephospho_pep, sequence) in batch:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
527 # do interesting stuff here...
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
528 PhosphoPep_UniProtSeq_LUT[phospho_pep] = phospho_pep
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
529 PhosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
530 (phospho_pep, DEPHOSPHOPEP)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
531 ] = dephospho_pep
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
532 if dephospho_pep not in DephosphoPep_UniProtSeq_LUT:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
533 DephosphoPep_UniProtSeq_LUT[dephospho_pep] = set()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
534 DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
535 (dephospho_pep, DESCRIPTION)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
536 ] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
537 DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
538 (dephospho_pep, GENE_NAME)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
539 ] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
540 DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
541 (dephospho_pep, UNIPROT_ID)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
542 ] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
543 DephosphoPep_UniProtSeq_LUT[(dephospho_pep, SEQUENCE)] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
544 DephosphoPep_UniProtSeq_LUT[dephospho_pep].add(phospho_pep)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
545
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
546 if (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
547 sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
548 not in DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
549 (dephospho_pep, SEQUENCE)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
550 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
551 ):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
552 DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
553 (dephospho_pep, SEQUENCE)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
554 ].append(sequence)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
555 for phospho_pep in DephosphoPep_UniProtSeq_LUT[dephospho_pep]:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
556 if phospho_pep != phospho_pep:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
557 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
558 "phospho_pep:'%s' phospho_pep:'%s'"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
559 % (phospho_pep, phospho_pep)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
560 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
561 if phospho_pep not in PhosphoPep_UniProtSeq_LUT:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
562 PhosphoPep_UniProtSeq_LUT[phospho_pep] = phospho_pep
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
563 PhosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
564 (phospho_pep, DEPHOSPHOPEP)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
565 ] = dephospho_pep
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
566 r = list(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
567 zip(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
568 [s for s in UniProtSeqLUT[(sequence, UNIPROT_ID)]],
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
569 [s for s in UniProtSeqLUT[(sequence, GENE_NAME)]],
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
570 [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
571 s
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
572 for s in UniProtSeqLUT[(sequence, DESCRIPTION)]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
573 ],
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
574 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
575 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
576 # Sort by `UniProt_ID`
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
577 # ref: https://stackoverflow.com/a/4174955/15509512
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
578 r = sorted(r, key=operator.itemgetter(0))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
579 # Get one tuple for each `phospho_pep`
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
580 # in DephosphoPep_UniProtSeq_LUT[dephospho_pep]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
581 for (upid, gn, desc) in r:
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
582 # Append pseudo-tuple per UniProt_ID iff not present
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
583 if (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
584 upid
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
585 not in DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
586 (dephospho_pep, UNIPROT_ID)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
587 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
588 ):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
589 DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
590 (dephospho_pep, UNIPROT_ID)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
591 ].append(upid)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
592 DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
593 (dephospho_pep, DESCRIPTION)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
594 ].append(desc)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
595 DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
596 (dephospho_pep, GENE_NAME)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
597 ].append(gn)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
598
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
599 # Close SwissProt SQLite database; clean up local variables
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
600 conn.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
601 # wipe local variables
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
602 phospho_pep = dephospho_pep = sequence = 0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
603 upid = gn = desc = r = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
604
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
605 # -------- Get SwissProt data from SQLite database (finish) -----------
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
606
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
607 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
608 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
609 "%0.6f finished reading and decoding '%s' [0.4]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
610 % (end_time - start_time, upstream_map_filename_tab),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
611 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
612 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
613
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
614 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
615 "{:>10} unique upstream phosphopeptides tested".format(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
616 str(len(upstream_map_p_peptide_list))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
617 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
618 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
619
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
620 # Read in Upstream tabular file
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
621 # We are discarding the intensity data; so read it as text
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
622 upstream_data = pandas.read_table(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
623 upstream_map_filename_tab, dtype="str", index_col=0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
624 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
625
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
626 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
627 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
628 "%0.6f read Upstream Map from file [1g_1]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
629 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
630 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
631 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
632
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
633 upstream_data.index = upstream_map_p_peptide_list
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
634
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
635 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
636 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
637 "%0.6f added index to Upstream Map [1g_2]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
638 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
639 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
640 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
641
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
642 # trim upstream_data to include only the upstream map columns
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
643 old_cols = upstream_data.columns.tolist()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
644 i = 0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
645 first_intensity = -1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
646 last_intensity = -1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
647 intensity_re = re.compile("Intensity.*")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
648 for col_name in old_cols:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
649 m = intensity_re.match(col_name)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
650 if m:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
651 last_intensity = i
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
652 if first_intensity == -1:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
653 first_intensity = i
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
654 i += 1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
655 # print('last intensity = %d' % last_intensity)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
656 col_PKCalpha = last_intensity + 2
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
657
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
658 data_in_cols = [old_cols[0]] + old_cols[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
659 first_intensity - 1: last_intensity
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
660 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
661 data_col_names = [old_cols[0]] + old_cols[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
662 first_intensity: last_intensity + 1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
663 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
664
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
665 if upstream_data.empty:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
666 print("upstream_data is empty")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
667 exit(0)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
668
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
669 data_in = upstream_data.copy(deep=True)[data_in_cols]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
670 data_in.columns = data_col_names
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
671 print("data_in")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
672 print(data_in)
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
673 ######################################################################
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
674
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
675 # Convert floating-point integers to int64 integers
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
676 # ref: https://stackoverflow.com/a/68497603/15509512
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
677 data_in[list(data_in.columns[1:])] = (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
678 data_in[list(data_in.columns[1:])]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
679 .astype("float64")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
680 .apply(np.int64)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
681 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
682
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
683 # create another phosphopeptide column that will be used to join later;
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
684 # MAY need to change depending on Phosphopeptide column position
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
685 # data_in[PHOSPHOPEPTIDE_MATCH] = data_in[data_in.columns.tolist()[0]]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
686 data_in[PHOSPHOPEPTIDE_MATCH] = data_in.index
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
687
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
688 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
689 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
690 "%0.6f set data_in[PHOSPHOPEPTIDE_MATCH] [A]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
691 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
692 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
693 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
694
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
695 # Produce a dictionary of metadata for a single phosphopeptide.
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
696 # This is replaces `UniProtInfo_subdict` from the original code.
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
697 def pseq_to_subdict(phospho_pep):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
698 # Strip "p" from phosphopeptide sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
699 dephospho_pep = re_phos.sub("", phospho_pep)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
700
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
701 # Determine number of phosphoresidues in phosphopeptide
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
702 numps = len(phospho_pep) - len(dephospho_pep)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
703
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
704 # Determine location(s) of phosphoresidue(s) in phosphopeptide
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
705 # (used later for Phosphoresidue, Sequence7, and Sequence10)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
706 ploc = [] # list of p locations
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
707 i = 0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
708 p = phospho_pep
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
709 while i < numps:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
710 ploc.append(p.find("p"))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
711 p = p[: p.find("p")] + p[p.find("p") + 1:]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
712 i += 1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
713
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
714 # Establish nested dictionary
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
715 result = {}
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
716 result[SEQUENCE] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
717 result[UNIPROT_ID] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
718 result[DESCRIPTION] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
719 result[GENE_NAME] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
720 result[PHOSPHORESIDUE] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
721 result[SEQUENCE7] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
722 result[SEQUENCE10] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
723
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
724 # Add stripped sequence to dictionary
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
725 result[SEQUENCE].append(dephospho_pep)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
726
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
727 # Locate phospho_pep in PhosphoPep_UniProtSeq_LUT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
728 # Caller may elect to:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
729 # try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
730 # ...
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
731 # except PreconditionError as pe:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
732 # print("'{expression}': {message}".format(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
733 # expression = pe.expression,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
734 # message = pe.message))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
735 # )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
736 # )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
737 if phospho_pep not in PhosphoPep_UniProtSeq_LUT:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
738 raise PreconditionError(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
739 phospho_pep,
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
740 " ".join([
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
741 "no matching phosphopeptide found in",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
742 "PhosphoPep_UniProtSeq_LUT"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
743 ])
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
744 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
745 if dephospho_pep not in DephosphoPep_UniProtSeq_LUT:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
746 raise PreconditionError(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
747 dephospho_pep,
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
748 " ".join([
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
749 "dephosphorylated phosphopeptide not found",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
750 "in DephosphoPep_UniProtSeq_LUT"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
751 ])
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
752 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
753 if (
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
754 dephospho_pep != PhosphoPep_UniProtSeq_LUT[
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
755 (phospho_pep, DEPHOSPHOPEP)]
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
756 ):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
757 my_err_msg = "dephosphorylated phosphopeptide does not match "
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
758 my_err_msg += "PhosphoPep_UniProtSeq_LUT"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
759 my_err_msg += "[(phospho_pep,DEPHOSPHOPEP)] = "
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
760 my_err_msg += PhosphoPep_UniProtSeq_LUT[
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
761 (phospho_pep, DEPHOSPHOPEP)]
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
762 raise PreconditionError(dephospho_pep, my_err_msg)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
763
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
764 result[SEQUENCE] = [dephospho_pep]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
765 result[UNIPROT_ID] = DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
766 (dephospho_pep, UNIPROT_ID)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
767 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
768 result[DESCRIPTION] = DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
769 (dephospho_pep, DESCRIPTION)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
770 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
771 result[GENE_NAME] = DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
772 (dephospho_pep, GENE_NAME)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
773 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
774 if (dephospho_pep, SEQUENCE) not in DephosphoPep_UniProtSeq_LUT:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
775 raise PreconditionError(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
776 dephospho_pep,
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
777 "".join(
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
778 "no matching phosphopeptide found",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
779 "in DephosphoPep_UniProtSeq_LUT")
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
780 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
781 UniProtSeqList = DephosphoPep_UniProtSeq_LUT[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
782 (dephospho_pep, SEQUENCE)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
783 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
784 if len(UniProtSeqList) < 1:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
785 print(
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
786 "Skipping DephosphoPep_UniProtSeq_LUT[('%s',SEQUENCE)] %s"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
787 % (dephospho_pep, "because value has zero length")
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
788 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
789 for UniProtSeq in UniProtSeqList:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
790 i = 0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
791 phosphoresidues = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
792 seq7s_set = set()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
793 seq7s = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
794 seq10s_set = set()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
795 seq10s = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
796 while i < len(ploc):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
797 start = UniProtSeq.find(dephospho_pep)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
798 # handle case where no sequence was found for dep-pep
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
799 if start < 0:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
800 i += 1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
801 continue
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
802 psite = (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
803 start + ploc[i]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
804 ) # location of phosphoresidue on protein sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
805
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
806 # add Phosphoresidue
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
807 phosphosite = "p" + str(UniProtSeq)[psite] + str(psite + 1)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
808 phosphoresidues.append(phosphosite)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
809
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
810 # Add Sequence7
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
811 if psite < 7: # phospho_pep at N terminus
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
812 seq7 = str(UniProtSeq)[: psite + 8]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
813 if seq7[psite] == "S": # if phosphosresidue is serine
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
814 pres = "s"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
815 elif (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
816 seq7[psite] == "T"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
817 ): # if phosphosresidue is threonine
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
818 pres = "t"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
819 elif (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
820 seq7[psite] == "Y"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
821 ): # if phosphoresidue is tyrosine
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
822 pres = "y"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
823 else: # if not pSTY
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
824 pres = "?"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
825 seq7 = (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
826 seq7[:psite] + pres + seq7[psite + 1: psite + 8]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
827 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
828 while (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
829 len(seq7) < 15
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
830 ): # add appropriate number of "_" to the front
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
831 seq7 = "_" + seq7
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
832 elif (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
833 len(UniProtSeq) - psite < 8
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
834 ): # phospho_pep at C terminus
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
835 seq7 = str(UniProtSeq)[psite - 7:]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
836 if seq7[7] == "S":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
837 pres = "s"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
838 elif seq7[7] == "T":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
839 pres = "t"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
840 elif seq7[7] == "Y":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
841 pres = "y"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
842 else:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
843 pres = "?"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
844 seq7 = seq7[:7] + pres + seq7[8:]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
845 while (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
846 len(seq7) < 15
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
847 ): # add appropriate number of "_" to the back
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
848 seq7 = seq7 + "_"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
849 else:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
850 seq7 = str(UniProtSeq)[psite - 7: psite + 8]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
851 pres = "" # phosphoresidue
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
852 if seq7[7] == "S": # if phosphosresidue is serine
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
853 pres = "s"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
854 elif seq7[7] == "T": # if phosphosresidue is threonine
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
855 pres = "t"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
856 elif seq7[7] == "Y": # if phosphoresidue is tyrosine
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
857 pres = "y"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
858 else: # if not pSTY
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
859 pres = "?"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
860 seq7 = seq7[:7] + pres + seq7[8:]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
861 if seq7 not in seq7s_set:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
862 seq7s.append(seq7)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
863 seq7s_set.add(seq7)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
864
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
865 # add Sequence10
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
866 if psite < 10: # phospho_pep at N terminus
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
867 seq10 = (
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
868 str(UniProtSeq)[:psite] + "p" + str(
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
869 UniProtSeq)[psite: psite + 11]
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
870 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
871 elif (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
872 len(UniProtSeq) - psite < 11
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
873 ): # phospho_pep at C terminus
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
874 seq10 = (
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
875 str(UniProtSeq)[psite - 10: psite] + "p" + str(
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
876 UniProtSeq)[psite:]
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
877 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
878 else:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
879 seq10 = str(UniProtSeq)[psite - 10: psite + 11]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
880 seq10 = seq10[:10] + "p" + seq10[10:]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
881 if seq10 not in seq10s_set:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
882 seq10s.append(seq10)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
883 seq10s_set.add(seq10)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
884
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
885 i += 1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
886
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
887 result[PHOSPHORESIDUE].append(phosphoresidues)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
888 result[SEQUENCE7].append(seq7s)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
889 # result[SEQUENCE10] is a list of lists of strings
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
890 result[SEQUENCE10].append(seq10s)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
891
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
892 r = list(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
893 zip(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
894 result[UNIPROT_ID],
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
895 result[GENE_NAME],
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
896 result[DESCRIPTION],
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
897 result[PHOSPHORESIDUE],
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
898 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
899 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
900 # Sort by `UniProt_ID`
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
901 # ref: https://stackoverflow.com//4174955/15509512
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
902 s = sorted(r, key=operator.itemgetter(0))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
903
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
904 result[UNIPROT_ID] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
905 result[GENE_NAME] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
906 result[DESCRIPTION] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
907 result[PHOSPHORESIDUE] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
908
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
909 for r in s:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
910 result[UNIPROT_ID].append(r[0])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
911 result[GENE_NAME].append(r[1])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
912 result[DESCRIPTION].append(r[2])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
913 result[PHOSPHORESIDUE].append(r[3])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
914
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
915 # convert lists to strings in the dictionary
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
916 for key, value in result.items():
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
917 if key not in [PHOSPHORESIDUE, SEQUENCE7, SEQUENCE10]:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
918 result[key] = "; ".join(map(str, value))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
919 elif key in [SEQUENCE10]:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
920 # result[SEQUENCE10] is a list of lists of strings
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
921 joined_value = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
922 joined_set = set()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
923 sep = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
924 for valL in value:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
925 # valL is a list of strings
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
926 for val in valL:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
927 # val is a string
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
928 if val not in joined_set:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
929 joined_set.add(val)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
930 joined_value += sep + val
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
931 sep = "; "
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
932 # joined_value is a string
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
933 result[key] = joined_value
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
934
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
935 newstring = "; ".join(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
936 [", ".join(prez) for prez in result[PHOSPHORESIDUE]]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
937 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
938 result[PHOSPHORESIDUE] = newstring
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
939
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
940 # separate sequence7's by |
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
941 oldstring = result[SEQUENCE7]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
942 oldlist = oldstring
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
943 newstring = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
944 for ol in oldlist:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
945 for e in ol:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
946 if e == ";":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
947 newstring = newstring + " |"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
948 elif len(newstring) > 0 and 1 > newstring.count(e):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
949 newstring = newstring + " | " + e
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
950 elif 1 > newstring.count(e):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
951 newstring = newstring + e
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
952 result[SEQUENCE7] = newstring
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
953
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
954 return [phospho_pep, result]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
955
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
956 # Construct list of [string, dictionary] lists
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
957 # where the dictionary provides the SwissProt metadata
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
958 # for a phosphopeptide
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
959 result_list = [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
960 whine(pseq_to_subdict, psequence)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
961 for psequence in data_in[PHOSPHOPEPTIDE_MATCH]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
962 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
963
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
964 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
965 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
966 "%0.6f added SwissProt annotations to phosphopeptides [B]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
967 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
968 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
969 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
970
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
971 # Construct dictionary from list of lists
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
972 # ref: https://www.8bitavenue.com/\
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
973 # how-to-convert-list-of-lists-to-dictionary-in-python/
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
974 UniProt_Info = {
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
975 result[0]: result[1]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
976 for result in result_list
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
977 if result is not None
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
978 }
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
979
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
980 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
981 print(
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
982 "%0.6f create dictionary mapping phosphopeptide %s"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
983 % (end_time - start_time, "to metadata dictionary [C]"),
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
984 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
985 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
986
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
987 # cosmetic: add N_A to phosphopeptide rows with no hits
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
988 p_peptide_list = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
989 for key in UniProt_Info:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
990 p_peptide_list.append(key)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
991 for nestedKey in UniProt_Info[key]:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
992 if UniProt_Info[key][nestedKey] == "":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
993 UniProt_Info[key][nestedKey] = N_A
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
994
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
995 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
996 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
997 "%0.6f performed cosmetic clean-up [D]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
998 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
999 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1000
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1001 # convert UniProt_Info dictionary to dataframe
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1002 uniprot_df = pandas.DataFrame.transpose(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1003 pandas.DataFrame.from_dict(UniProt_Info)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1004 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1005
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1006 # reorder columns to match expected output file
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1007 uniprot_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1008 PHOSPHOPEPTIDE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1009 ] = uniprot_df.index # make index a column too
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1010
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1011 cols = uniprot_df.columns.tolist()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1012 # cols = [cols[-1]]+cols[4:6]+[cols[1]]+[cols[2]]+[cols[6]]+[cols[0]]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1013 # uniprot_df = uniprot_df[cols]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1014 uniprot_df = uniprot_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1015 [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1016 PHOSPHOPEPTIDE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1017 SEQUENCE10,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1018 SEQUENCE7,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1019 GENE_NAME,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1020 PHOSPHORESIDUE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1021 UNIPROT_ID,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1022 DESCRIPTION,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1023 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1024 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1025
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1026 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1027 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1028 "%0.6f reordered columns to match expected output file [1]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1029 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1030 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1031 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1032
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1033 # concat to split then groupby to collapse
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1034 seq7_df = pandas.concat(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1035 [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1036 pandas.Series(row[PHOSPHOPEPTIDE], row[SEQUENCE7].split(" | "))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1037 for _, row in uniprot_df.iterrows()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1038 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1039 ).reset_index()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1040 seq7_df.columns = [SEQUENCE7, PHOSPHOPEPTIDE]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1041
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1042 # read in PhosphoSitePlus Regulatory Sites dataset from SQLite
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1043 # --- -------------- begin read PSP_Regulatory_sites -----
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1044 conn = sql.connect(uniprot_sqlite)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1045 regsites_df = pandas.read_sql_query(PSP_REGSITE_SQL, conn)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1046 # Close SwissProt SQLite database
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1047 conn.close()
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1048 # ... -------------- end read PSP_Regulatory_sites -------
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1049
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1050 # keep only the human entries in dataframe
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1051 if len(species) > 0:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1052 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1053 'Limit PhosphoSitesPlus records to species "' + species + '"'
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1054 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1055 regsites_df = regsites_df[regsites_df.ORGANISM == species]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1056
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1057 # merge the seq7 df with the regsites df based off of the sequence7
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1058 merge_df = seq7_df.merge(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1059 regsites_df,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1060 left_on=SEQUENCE7,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1061 right_on=SITE_PLUSMINUS_7AA_SQL,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1062 how="left",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1063 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1064
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1065 # after merging df, select only the columns of interest;
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1066 # note that PROTEIN is absent here
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1067 merge_df = merge_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1068 [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1069 PHOSPHOPEPTIDE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1070 SEQUENCE7,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1071 ON_FUNCTION,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1072 ON_PROCESS,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1073 ON_PROT_INTERACT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1074 ON_OTHER_INTERACT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1075 ON_NOTES,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1076 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1077 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1078 # combine column values of interest
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1079 # into one FUNCTION_PHOSPHORESIDUE column"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1080 merge_df[FUNCTION_PHOSPHORESIDUE] = merge_df[ON_FUNCTION].str.cat(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1081 merge_df[ON_PROCESS], sep="; ", na_rep=""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1082 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1083 merge_df[FUNCTION_PHOSPHORESIDUE] = merge_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1084 FUNCTION_PHOSPHORESIDUE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1085 ].str.cat(merge_df[ON_PROT_INTERACT], sep="; ", na_rep="")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1086 merge_df[FUNCTION_PHOSPHORESIDUE] = merge_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1087 FUNCTION_PHOSPHORESIDUE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1088 ].str.cat(merge_df[ON_OTHER_INTERACT], sep="; ", na_rep="")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1089 merge_df[FUNCTION_PHOSPHORESIDUE] = merge_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1090 FUNCTION_PHOSPHORESIDUE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1091 ].str.cat(merge_df[ON_NOTES], sep="; ", na_rep="")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1092
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1093 # remove the columns that were combined
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1094 merge_df = merge_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1095 [PHOSPHOPEPTIDE, SEQUENCE7, FUNCTION_PHOSPHORESIDUE]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1096 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1097
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1098 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1099 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1100 "%0.6f merge regsite metadata [1a]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1101 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1102 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1103
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1104 # cosmetic changes to Function Phosphoresidue column
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1105 fp_series = pandas.Series(merge_df[FUNCTION_PHOSPHORESIDUE])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1106
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1107 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1108 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1109 "%0.6f more cosmetic changes [1b]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1110 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1111 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1112
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1113 i = 0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1114 while i < len(fp_series):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1115 # remove the extra ";" so that it looks more professional
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1116 if fp_series[i] == "; ; ; ; ": # remove ; from empty hits
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1117 fp_series[i] = ""
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1118 while fp_series[i].endswith("; "): # remove ; from the ends
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1119 fp_series[i] = fp_series[i][:-2]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1120 while fp_series[i].startswith("; "): # remove ; from the beginning
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1121 fp_series[i] = fp_series[i][2:]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1122 fp_series[i] = fp_series[i].replace("; ; ; ; ", "; ")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1123 fp_series[i] = fp_series[i].replace("; ; ; ", "; ")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1124 fp_series[i] = fp_series[i].replace("; ; ", "; ")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1125
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1126 # turn blanks into N_A to signify the info
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1127 # that was searched for but cannot be found
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1128 if fp_series[i] == "":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1129 fp_series[i] = N_A
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1130
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1131 i += 1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1132 merge_df[FUNCTION_PHOSPHORESIDUE] = fp_series
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1133
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1134 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1135 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1136 "%0.6f cleaned up semicolons [1c]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1137 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1138 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1139
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1140 # merge uniprot df with merge df
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1141 uniprot_regsites_merged_df = uniprot_df.merge(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1142 merge_df,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1143 left_on=PHOSPHOPEPTIDE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1144 right_on=PHOSPHOPEPTIDE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1145 how="left",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1146 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1147
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1148 # collapse the merged df
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1149 uniprot_regsites_collapsed_df = pandas.DataFrame(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1150 uniprot_regsites_merged_df.groupby(PHOSPHOPEPTIDE)[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1151 FUNCTION_PHOSPHORESIDUE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1152 ].apply(lambda x: ppep_join(x))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1153 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1154 # .apply(lambda x: "%s" % ' | '.join(x)))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1155
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1156 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1157 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1158 "%0.6f collapsed pandas dataframe [1d]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1159 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1160 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1161
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1162 uniprot_regsites_collapsed_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1163 PHOSPHOPEPTIDE
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1164 ] = (
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1165 uniprot_regsites_collapsed_df.index
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1166 ) # add df index as its own column
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1167
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1168 # rename columns
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1169 uniprot_regsites_collapsed_df.columns = [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1170 FUNCTION_PHOSPHORESIDUE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1171 "ppp",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1172 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1173
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1174 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1175 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1176 "%0.6f selected columns to be merged to uniprot_df [1e]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1177 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1178 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1179 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1180
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1181 # add columns based on Sequence7 matching site_+/-7_AA
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1182 uniprot_regsite_df = pandas.merge(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1183 left=uniprot_df,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1184 right=uniprot_regsites_collapsed_df,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1185 how="left",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1186 left_on=PHOSPHOPEPTIDE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1187 right_on="ppp",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1188 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1189
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1190 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1191 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1192 "%0.6f added columns based on Sequence7 matching site_+/-7_AA [1f]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1193 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1194 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1195 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1196
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1197 data_in.rename(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1198 {"Protein description": PHOSPHOPEPTIDE},
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1199 axis="columns",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1200 inplace=True,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1201 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1202
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1203 res2 = sorted(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1204 data_in[PHOSPHOPEPTIDE_MATCH].tolist(), key=lambda s: s.casefold()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1205 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1206 data_in = data_in.loc[res2]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1207
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1208 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1209 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1210 "%0.6f sorting time [1f]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1211 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1212 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1213
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1214 print("old_cols[:col_PKCalpha]")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1215 print(old_cols[:col_PKCalpha])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1216 cols = [old_cols[0]] + old_cols[col_PKCalpha - 1:]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1217 upstream_data = upstream_data[cols]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1218 print("upstream_data.columns")
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1219 print(upstream_data.columns)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1220
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1221 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1222 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1223 "%0.6f refactored columns for Upstream Map [1g]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1224 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1225 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1226 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1227
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1228 # rename upstream columns in new list
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1229 def col_rename(name):
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1230 if "_NetworKIN" in name:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1231 name = name.split("_")[0]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1232 if " motif" in name:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1233 name = name.split(" motif")[0]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1234 if " sequence " in name:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1235 name = name.split(" sequence")[0]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1236 if "_Phosida" in name:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1237 name = name.split("_")[0]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1238 if "_PhosphoSite" in name:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1239 name = name.split("_")[0]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1240 return name
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1241
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1242 new_cols = [col_rename(col) for col in cols]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1243 upstream_data.columns = new_cols
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1244
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1245 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1246 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1247 "%0.6f renamed columns for Upstream Map [1h_1]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1248 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1249 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1250 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1251
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1252 # Create upstream_data_cast as a copy of upstream_data
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1253 # but with first column substituted by the phosphopeptide sequence
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1254 upstream_data_cast = upstream_data.copy()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1255 new_cols_cast = new_cols
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1256 new_cols_cast[0] = "p_peptide"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1257 upstream_data_cast.columns = new_cols_cast
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1258 upstream_data_cast["p_peptide"] = upstream_data.index
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1259
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1260 # Get melted kinase mapping data from SQLite database
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1261 # --- begin read upstream_data_melt ---------------------------------
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1262 conn = sql.connect(uniprot_sqlite)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1263 upstream_data_melt_df = pandas.read_sql_query(PPEP_MELT_SQL, conn)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1264 # Close SwissProt SQLite database
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1265 conn.close()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1266 upstream_data_melt = upstream_data_melt_df.copy()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1267 upstream_data_melt.columns = ["p_peptide", "characterization", "X"]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1268 upstream_data_melt["characterization"] = [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1269 col_rename(s) for s in upstream_data_melt["characterization"]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1270 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1271
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1272 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1273 "%0.6f upstream_data_melt_df initially has %d rows"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1274 % (end_time - start_time, len(upstream_data_melt.axes[0])),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1275 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1276 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1277 # ref: https://stackoverflow.com/a/27360130/15509512
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1278 # e.g. df.drop(df[df.score < 50].index, inplace=True)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1279 upstream_data_melt.drop(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1280 upstream_data_melt[upstream_data_melt.X != "X"].index, inplace=True
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1281 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1282 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1283 "%0.6f upstream_data_melt_df pre-dedup has %d rows"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1284 % (end_time - start_time, len(upstream_data_melt.axes[0])),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1285 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1286 )
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1287 # ... end read upstream_data_melt ---------------------------------
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1288
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1289 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1290 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1291 "%0.6f melted and minimized Upstream Map dataframe [1h_2]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1292 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1293 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1294 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1295 # ... end read upstream_data_melt
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1296
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1297 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1298 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1299 "%0.6f indexed melted Upstream Map [1h_2a]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1300 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1301 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1302 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1303
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1304 upstream_delta_melt_LoL = upstream_data_melt.values.tolist()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1305
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1306 melt_dict = {}
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1307 for key in upstream_map_p_peptide_list:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1308 melt_dict[key] = []
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1309
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1310 for el in upstream_delta_melt_LoL:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1311 (p_peptide, characterization, X) = tuple(el)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1312 if p_peptide in melt_dict:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1313 melt_dict[p_peptide].append(characterization)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1314 else:
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1315 los = [
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1316 "Phosphopeptide %s" % p_peptide,
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1317 "not found in ppep_mapping_db:",
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1318 '"phopsphopeptides" and "ppep_mapping_db" must both',
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1319 "originate from the same run of mqppep_kinase_mapping"
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1320 ]
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1321 exit(" ".join(los))
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1322
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1323 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1324 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1325 "%0.6f appended peptide characterizations [1h_2b]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1326 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1327 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1328 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1329
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1330 # for key in upstream_map_p_peptide_list:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1331 # melt_dict[key] = ' | '.join(melt_dict[key])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1332
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1333 for key in upstream_map_p_peptide_list:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1334 melt_dict[key] = melt_join(melt_dict[key])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1335
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1336 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1337 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1338 "%0.6f concatenated multiple characterizations [1h_2c]"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1339 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1340 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1341 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1342
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1343 # map_dict is a dictionary of dictionaries
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1344 map_dict = {}
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1345 for key in upstream_map_p_peptide_list:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1346 map_dict[key] = {}
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1347 map_dict[key][PUTATIVE_UPSTREAM_DOMAINS] = melt_dict[key]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1348
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1349 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1350 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1351 "%0.6f instantiated map dictionary [2]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1352 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1353 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1354
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1355 # convert map_dict to dataframe
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1356 map_df = pandas.DataFrame.transpose(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1357 pandas.DataFrame.from_dict(map_dict)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1358 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1359 map_df["p-peptide"] = map_df.index # make index a column too
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1360 cols_map_df = map_df.columns.tolist()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1361 cols_map_df = [cols_map_df[1]] + [cols_map_df[0]]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1362 map_df = map_df[cols_map_df]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1363
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1364 # join map_df to uniprot_regsite_df
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1365 output_df = uniprot_regsite_df.merge(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1366 map_df, how="left", left_on=PHOSPHOPEPTIDE, right_on="p-peptide"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1367 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1368
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1369 output_df = output_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1370 [
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1371 PHOSPHOPEPTIDE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1372 SEQUENCE10,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1373 SEQUENCE7,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1374 GENE_NAME,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1375 PHOSPHORESIDUE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1376 UNIPROT_ID,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1377 DESCRIPTION,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1378 FUNCTION_PHOSPHORESIDUE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1379 PUTATIVE_UPSTREAM_DOMAINS,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1380 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1381 ]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1382
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1383 # join output_df back to quantitative columns in data_in df
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1384 quant_cols = data_in.columns.tolist()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1385 quant_cols = quant_cols[1:]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1386 quant_data = data_in[quant_cols]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1387
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1388 # ---- Write merge/filter metadata to SQLite database (start) ----
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1389 # Open SwissProt SQLite database
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1390 conn = sql.connect(output_sqlite)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1391 cur = conn.cursor()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1392
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1393 cur.executescript(MRGFLTR_DDL)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1394
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1395 cur.execute(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1396 CITATION_INSERT_STMT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1397 ("mrgfltr_metadata_view", CITATION_INSERT_PSP),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1398 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1399 cur.execute(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1400 CITATION_INSERT_STMT, ("mrgfltr_metadata", CITATION_INSERT_PSP)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1401 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1402 cur.execute(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1403 CITATION_INSERT_STMT,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1404 ("mrgfltr_metadata_view", CITATION_INSERT_PSP_REF),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1405 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1406 cur.execute(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1407 CITATION_INSERT_STMT, ("mrgfltr_metadata", CITATION_INSERT_PSP_REF)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1408 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1409
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1410 # Read ppep-to-sequence LUT
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1411 ppep_lut_df = pandas.read_sql_query(PPEP_ID_SQL, conn)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1412 # write only metadata for merged/filtered records to SQLite
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1413 mrgfltr_metadata_df = output_df.copy()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1414 # replace phosphopeptide seq with ppep.id
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1415 mrgfltr_metadata_df = ppep_lut_df.merge(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1416 mrgfltr_metadata_df,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1417 left_on="ppep_seq",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1418 right_on=PHOSPHOPEPTIDE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1419 how="inner",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1420 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1421 mrgfltr_metadata_df.drop(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1422 columns=[PHOSPHOPEPTIDE, "ppep_seq"], inplace=True
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1423 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1424 # rename columns
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1425 mrgfltr_metadata_df.columns = MRGFLTR_METADATA_COLUMNS
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1426 mrgfltr_metadata_df.to_sql(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1427 "mrgfltr_metadata",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1428 con=conn,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1429 if_exists="append",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1430 index=False,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1431 method="multi",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1432 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1433
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1434 # Close SwissProt SQLite database
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1435 conn.close()
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1436 # ---- Write merge/filter metadata to SQLite database (finish) ----
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1437
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1438 output_df = output_df.merge(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1439 quant_data,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1440 how="right",
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1441 left_on=PHOSPHOPEPTIDE,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1442 right_on=PHOSPHOPEPTIDE_MATCH,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1443 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1444 output_cols = output_df.columns.tolist()
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1445 output_cols = output_cols[:-1]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1446 output_df = output_df[output_cols]
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1447
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1448 # cosmetic changes to Upstream column
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1449 # fill the NaN with "" for those Phosphopeptides that got a
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1450 # "WARNING: Failed match for " in the upstream mapping
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1451 output_df[PUTATIVE_UPSTREAM_DOMAINS] = output_df[
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1452 PUTATIVE_UPSTREAM_DOMAINS
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1453 ].fillna(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1454 ""
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1455 )
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1456 us_series = pandas.Series(output_df[PUTATIVE_UPSTREAM_DOMAINS])
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1457 i = 0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1458 while i < len(us_series):
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1459 # turn blanks into N_A to signify the info
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1460 # that was searched for but cannot be found
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1461 if us_series[i] == "":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1462 us_series[i] = N_A
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1463 i += 1
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1464 output_df[PUTATIVE_UPSTREAM_DOMAINS] = us_series
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1465
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1466 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1467 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1468 "%0.6f establisheed output [3]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1469 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1470 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1471
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1472 (output_rows, output_cols) = output_df.shape
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1473
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1474 output_df = output_df.convert_dtypes(convert_integer=True)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1475
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1476 # Output onto Final CSV file
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1477 output_df.to_csv(output_filename_csv, index=False)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1478 output_df.to_csv(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1479 output_filename_tab, quoting=None, sep="\t", index=False
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1480 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1481
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1482 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1483 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1484 "%0.6f wrote output [4]" % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1485 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1486 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1487
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1488 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1489 "{:>10} phosphopeptides written to output".format(str(output_rows))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1490 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1491
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1492 end_time = time.process_time() # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1493 print(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1494 "%0.6f seconds of non-system CPU time were consumed"
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1495 % (end_time - start_time,),
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1496 file=sys.stderr,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1497 ) # timer
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1498
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1499 # Rev. 7/1/2016
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1500 # Rev. 7/3/2016 : fill NaN in Upstream column to replace to N/A's
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1501 # Rev. 7/3/2016: renamed Upstream column to PUTATIVE_UPSTREAM_DOMAINS
1
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1502 # Rev. 12/2/2021: Converted to Python from ipynb; use fast \
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1503 # Aho-Corasick searching; \
b76c75521d91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 43e7a43b545c24b2dc33d039198551c032aa79be
galaxyp
parents: 0
diff changeset
1504 # read from SwissProt SQLite database
0
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1505 # Rev. 12/9/2021: Transfer code to Galaxy tool wrapper
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1506
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1507 #
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1508 # copied from Excel Output Script.ipynb END #
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1509 #
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1510
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1511 try:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1512 catch(
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1513 mqpep_getswissprot,
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1514 )
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1515 exit(0)
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1516 except Exception as e:
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1517 exit("Internal error running mqpep_getswissprot(): %s" % (e))
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1518
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1519
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1520 if __name__ == "__main__":
8dfd5d2b5903 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1521 __main__()