annotate rbpfeatures.py @ 3:ecf125a1ad73 draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit fb1fc1e3a062adfb1a78cd46aafd6a6e9f1e3bab
author rnateam
date Tue, 19 Jul 2016 10:22:02 -0400
parents 8918de535391
children a609d6dc8047
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
1 """Compute the RBP features."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
2
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
3 import re
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
4 import subprocess as sp
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
5 import uuid
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
6 from os import mkdir
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
7 from os import listdir
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
8 from os.path import isfile, join
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
9 from os import devnull
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
10 from shutil import rmtree
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
11
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
12 import numpy as np
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
13
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
14 import pandas as pd
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
15
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
16 import fasta_utils
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
17 import pfam_utils
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
18
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
19 __author__ = "Gianluca Corrado"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
20 __copyright__ = "Copyright 2016, Gianluca Corrado"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
21 __license__ = "MIT"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
22 __maintainer__ = "Gianluca Corrado"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
23 __email__ = "gianluca.corrado@unitn.it"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
24 __status__ = "Production"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
25
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
26
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
27 class RBPVectorizer():
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
28 """Compute the RBP features."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
29
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
30 def __init__(self, fasta):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
31 """
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
32 Constructor.
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
33
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
34 Parameters
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
35 ----------
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
36 fasta : str
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
37 Fasta file containing the RBP sequences to predict.
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
38 """
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
39 self.fasta = fasta
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
40
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
41 self._mod_fold = "AURA_Human_data/RBP_features/mod"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
42 self._reference_fisher_scores = \
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
43 "AURA_Human_data/RBP_features/fisher_scores_ref"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
44 self._train_rbps_file = \
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
45 "AURA_Human_data/RBP_features/rbps_in_train.txt"
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
46
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
47 self._temp_fold = "temp_" + str(uuid.uuid4())
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
48 self.pfam_scan = "%s/pfam_scan.txt" % self._temp_fold
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
49 self._dom_fold = "%s/domains" % self._temp_fold
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
50 self._seeds_fold = "%s/seeds" % self._temp_fold
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
51 self._fisher_fold = "%s/fisher_scores" % self._temp_fold
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
52
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
53 def _pfam_scan(self):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
54 """Scan the sequences against the Pfam database."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
55 nf = open(self.pfam_scan, "w")
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
56 nf.write(pfam_utils.search_header())
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
57
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
58 fasta = fasta_utils.import_fasta(self.fasta)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
59
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
60 for rbp in sorted(fasta.keys()):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
61 seq = fasta[rbp]
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
62 text = pfam_utils.sequence_search(rbp, seq)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
63 nf.write(text)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
64
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
65 nf.close()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
66
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
67 def _overlapping_domains(self):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
68 """Compute the set of domains contributing to the similarity."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
69 reference_domains = set([dom.replace(".mod", "") for dom in
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
70 listdir(self._mod_fold) if
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
71 isfile(join(self._mod_fold, dom))])
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
72
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
73 data = pfam_utils.read_pfam_output(self.pfam_scan)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
74 if data is None:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
75 return []
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
76
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
77 prot_domains = set([a.split('.')[0] for a in data["hmm_acc"]])
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
78 dom_list = sorted(list(reference_domains & prot_domains))
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
79
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
80 return dom_list
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
81
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
82 def _prepare_domains(self, dom_list):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
83 """Select domain subsequences from the entire protein sequences."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
84 def prepare_domains(fasta_dic, dom_list, pfam_scan, out_folder):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
85 out_file_dic = {}
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
86 for acc in dom_list:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
87 out_file_dic[acc] = open("%s/%s.fa" % (out_folder, acc), "w")
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
88
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
89 f = open(pfam_scan)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
90 f.readline()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
91 for line in f:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
92 split = line.split()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
93 rbp = split[0]
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
94 start = int(split[3])
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
95 stop = int(split[4])
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
96 acc = split[5].split('.')[0]
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
97 if acc in out_file_dic.keys():
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
98 out_file_dic[acc].write(
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
99 ">%s:%i-%i\n%s\n" % (rbp, start, stop,
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
100 fasta_dic[rbp][start:stop]))
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
101 f.close()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
102
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
103 for acc in dom_list:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
104 out_file_dic[acc].close()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
105
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
106 mkdir(self._dom_fold)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
107 fasta = fasta_utils.import_fasta(self.fasta)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
108 prepare_domains(fasta, dom_list, self.pfam_scan,
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
109 self._dom_fold)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
110
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
111 def _compute_fisher_scores(self, dom_list):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
112 """Wrapper for SAM 3.5 get_fisher_scores."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
113 def get_fisher_scores(dom_list, mod_fold, dom_fold, fisher_fold):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
114 for acc in dom_list:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
115 _FNULL = open(devnull, 'w')
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
116 cmd = "get_fisher_scores run -i %s/%s.mod -db %s/%s.fa" % (
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
117 mod_fold, acc, dom_fold, acc)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
118 fisher = sp.check_output(
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
119 cmd, shell=True, stderr=_FNULL)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
120 nf = open("%s/%s.txt" % (fisher_fold, acc), "w")
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
121 nf.write(fisher)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
122 nf.close()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
123
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
124 mkdir(self._fisher_fold)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
125 get_fisher_scores(dom_list, self._mod_fold, self._dom_fold,
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
126 self._fisher_fold)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
127
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
128 def _ekm(self, dom_list):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
129 """Compute the empirical kernel map from the Fisher scores."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
130 def process_seg(e):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
131 """Process segment of a SAM 3.5 get_fisher_scores output file."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
132 seg = e.split()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
133 c = seg[0].split(':')[0]
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
134 m = map(float, seg[3:])
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
135 return c, m
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
136
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
137 def read_sam_file(samfile):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
138 """Read a SAM 3.5 get_fisher_scores output file."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
139 f = open(samfile)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
140 data = f.read()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
141 f.close()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
142
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
143 columns = []
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
144 m = []
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
145 split = re.split(">A ", data)[1:]
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
146 for e in split:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
147 c, m_ = process_seg(e)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
148 columns.append(c)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
149 m.append(m_)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
150
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
151 m = np.matrix(m)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
152 df = pd.DataFrame(data=m.T, columns=columns)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
153 return df
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
154
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
155 def dom_features(fisher_fold, dom_list, names=None):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
156 """Compute the features with respect to a domain type."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
157 dfs = []
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
158 for acc in dom_list:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
159 df = read_sam_file("%s/%s.txt" % (fisher_fold, acc))
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
160 df = df.groupby(df.columns, axis=1).mean()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
161 dfs.append(df)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
162
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
163 con = pd.concat(dfs, ignore_index=True)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
164
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
165 if names is not None:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
166 add = sorted(list(set(names) - set(con.columns)))
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
167 fil = sorted(list(set(names) - set(add)))
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
168 con = con[fil]
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
169 for c in add:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
170 con[c] = np.zeros(len(con.index), dtype='float64')
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
171 con = con[names]
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
172
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
173 con = con.fillna(0.0)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
174 return con
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
175
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
176 f = open(self._train_rbps_file)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
177 train_rbps = f.read().strip().split('\n')
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
178 f.close()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
179 ref = dom_features(self._reference_fisher_scores, dom_list,
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
180 names=train_rbps)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
181 ekm_ref = ref.T.dot(ref)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
182 ekm_ref.index = ekm_ref.columns
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
183
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
184 sel = dom_features(self._fisher_fold, dom_list)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
185
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
186 ekm_sel = sel.T.dot(sel)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
187 ekm_sel.index = ekm_sel.columns
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
188
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
189 ekm = ref.T.dot(sel)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
190
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
191 for rs in ekm.columns:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
192 for rr in ekm.index:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
193 if ekm_ref[rr][rr] > 0 and ekm_sel[rs][rs] > 0:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
194 ekm[rs][rr] /= np.sqrt(ekm_ref[rr][rr] * ekm_sel[rs][rs])
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
195 return ekm
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
196
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
197 def vectorize(self):
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
198 """Produce the RBP features."""
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
199 # create a temporary folder
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
200 mkdir(self._temp_fold)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
201 # scan the RBP sequences against Pfam
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
202 self._pfam_scan()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
203 # determine the accession numbers of the pfam domains needed for
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
204 # computing the features
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
205 dom_list = self._overlapping_domains()
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
206 if len(dom_list) == 0:
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
207 rmtree(self._temp_fold)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
208 return None
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
209 # prepare fasta file with the sequence of the domains
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
210 self._prepare_domains(dom_list)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
211 # compute fisher scores using SAM 3.5
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
212 self._compute_fisher_scores(dom_list)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
213 # compute the empirical kernel map
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
214 ekm = self._ekm(dom_list)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
215 # remove the temporary folder
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
216 rmtree(self._temp_fold)
8918de535391 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 2fc7f3c08f30e2d81dc4ad19759dfe7ba9b0a3a1
rnateam
parents:
diff changeset
217 return ekm