Mercurial > repos > artbio > blast_unmatched

annotate blast_unmatched.py @ 2:dfcdac284538 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 830e10a94c2afc178f4078609842cd93808df1b4
author artbio
date Thu, 05 Oct 2017 05:11:01 -0400
parents 50c1fa95a076
children fffdb903f2d1
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
1 #!/usr/bin/env python3
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
2
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
3 import optparse
2
dfcdac284538 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 830e10a94c2afc178f4078609842cd93808df1b4
artbio
parents: 1
diff changeset
4 import re
0
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
5
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
6 def parse_options():
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
7 """
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
8 Parse the options guiven to the script
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
9 """
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
10 parser = optparse.OptionParser(description='Get unmatched blast queries')
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
11 parser.add_option('-f','--fasta', dest='fasta_file', help='Query fasta file\
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
12 used during blast')
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
13 parser.add_option('-b','--blast', dest='blast_file', help='Blast tabular\
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
14 output (queries in 1rst column)')
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
15 parser.add_option('-o','--output', dest='output_file', help='Output file name')
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
16 (options, args) = parser.parse_args()
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
17 if len(args) > 0:
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
18 parser.error('Wrong number of arguments')
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
19 return options
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
20
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
21 def get_matched(blast_file):
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
22 """
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
23 Get a dictionary of all the queries that got a match
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
24 """
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
25 matched = dict()
1
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
26 with open(blast_file, 'r') as infile:
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
27 for line in infile:
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
28 fields = line.split("\t")
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
29 query_id = fields[0]
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
30 matched[query_id] = 1
0
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
31 return matched
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
32
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
33 def get_unmatched(output_file, fasta_file, matched):
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
34 """
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
35 Compares matched queries to query fasta file and print unmatched to ouput
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
36 """
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
37 output_file_handle = open(output_file, 'w')
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
38 unmatched = False
2
dfcdac284538 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 830e10a94c2afc178f4078609842cd93808df1b4
artbio
parents: 1
diff changeset
39 end = re.compile(".+\W$")
1
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
40 with open(fasta_file, 'r') as infile:
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
41 for line in infile:
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
42 if line.startswith('>'):
2
dfcdac284538 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 830e10a94c2afc178f4078609842cd93808df1b4
artbio
parents: 1
diff changeset
43 subline = line[1:].rstrip() #qid are 100chars long in blast
dfcdac284538 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 830e10a94c2afc178f4078609842cd93808df1b4
artbio
parents: 1
diff changeset
44 while end.match(subline) != None:
dfcdac284538 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 830e10a94c2afc178f4078609842cd93808df1b4
artbio
parents: 1
diff changeset
45 subline = subline[:-1]
1
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
46 if subline not in matched:
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
47 output_file_handle.write(line)
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
48 unmatched = True
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
49 else:
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
50 unmatched = False
50c1fa95a076 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit be082f72c8d8c1eebe3f5643da1a73ab0ac9e4b3
artbio
parents: 0
diff changeset
51 elif unmatched:
0
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
52 output_file_handle.write(line)
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
53 output_file_handle.close()
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
54
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
55 def __main__():
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
56 opts = parse_options()
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
57 matched = get_matched(opts.blast_file)
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
58 get_unmatched(opts.output_file, opts.fasta_file, matched)
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
59
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
60 if __main__():
f3b63b59a1ea planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/blast_unmatched commit 5bd801feb838592fbb1f6dd68b5f1a480042da40
artbio
parents:
diff changeset
61 __main__()