Mercurial > repos > earlhaminst > t_coffee
annotate filter_by_fasta_ids.py @ 0:794a6e864a96 draft
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
author | earlhaminst |
---|---|
date | Thu, 15 Dec 2016 11:04:25 -0500 |
parents | |
children | b3833e5b50d4 |
rev | line source |
---|---|
0
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
1 #!/usr/bin/env python |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
2 """ A script to build specific fasta databases """ |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
3 from __future__ import print_function |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
4 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
5 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
6 import logging |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
7 import sys |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
8 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
9 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
10 # ===================================== Iterator =============================== |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
11 class Sequence: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
12 ''' Holds protein sequence information ''' |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
13 def __init__(self): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
14 self.header = "" |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
15 self.sequence_parts = [] |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
16 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
17 def get_sequence(self): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
18 return "".join([line.rstrip().replace('\n', '').replace('\r', '') for line in self.sequence_parts]) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
19 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
20 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
21 class FASTAReader: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
22 """ |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
23 FASTA db iterator. Returns a single FASTA sequence object. |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
24 """ |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
25 def __init__(self, fasta_name): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
26 self.fasta_file = open(fasta_name) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
27 self.next_line = self.fasta_file.readline() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
28 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
29 def __iter__(self): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
30 return self |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
31 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
32 def __next__(self): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
33 ''' Iteration ''' |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
34 # while True: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
35 # line = self.fasta_file.readline() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
36 # if not line: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
37 # raise StopIteration |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
38 # if line[0] == '>': |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
39 # break |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
40 next_line = self.next_line |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
41 if not next_line: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
42 raise StopIteration |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
43 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
44 seq = Sequence() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
45 seq.header = next_line.rstrip().replace('\n', '').replace('\r', '') |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
46 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
47 next_line = self.fasta_file.readline() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
48 while next_line and next_line[0] != '>': |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
49 # tail = self.fasta_file.tell() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
50 # line = self.fasta_file.readline() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
51 # if not line: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
52 # break |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
53 # if line[0] == '>': |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
54 # self.fasta_file.seek(tail) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
55 # break |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
56 seq.sequence_parts.append(next_line) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
57 next_line = self.fasta_file.readline() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
58 self.next_line = next_line |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
59 return seq |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
60 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
61 # Python 2/3 compat |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
62 next = __next__ |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
63 # ============================================================================== |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
64 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
65 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
66 def target_match(target, search_entry): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
67 ''' Matches ''' |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
68 search_entry = search_entry.upper() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
69 for atarget in target: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
70 if search_entry.find(atarget) > -1: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
71 return atarget |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
72 return None |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
73 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
74 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
75 def main(): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
76 ''' the main function''' |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
77 logging.basicConfig(filename='filter_fasta_log', |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
78 level=logging.INFO, |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
79 format='%(asctime)s :: %(levelname)s :: %(message)s',) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
80 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
81 used_sequences = set() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
82 work_summary = {'wanted': 0, 'found': 0, 'duplicates': 0} |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
83 targets = [] |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
84 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
85 f_target = open(sys.argv[1]) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
86 for line in f_target.readlines(): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
87 targets.append(">%s" % line.strip().upper()) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
88 f_target.close() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
89 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
90 work_summary['wanted'] = len(targets) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
91 homd_db = FASTAReader(sys.argv[2]) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
92 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
93 # output = open(sys.argv[3], "w") |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
94 for entry in homd_db: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
95 target_matched_results = target_match(targets, entry.header) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
96 if target_matched_results: |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
97 work_summary['found'] += 1 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
98 targets.remove(target_matched_results) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
99 sequence = entry.get_sequence() |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
100 used_sequences.add(sequence) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
101 print(entry.header) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
102 print(sequence) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
103 for parm, count in work_summary.items(): |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
104 logging.info('%s ==> %d', parm, count) |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
105 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
106 |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
107 if __name__ == "__main__": |
794a6e864a96
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/t_coffee commit 230ae552ddeb1bfdef3a09becaa5c6d373529a05-dirty
earlhaminst
parents:
diff
changeset
|
108 main() |