Mercurial > repos > earlhaminst > blast_parser

annotate blast_parser.py @ 3:70df762b48a8 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
author earlhaminst
date Tue, 03 Oct 2017 04:51:45 -0400
parents
children 363f3480622d
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
1 """
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
2 Simple parser to convert a BLAST 12-column or 24-column tabular output into a
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
3 3-column tabular input for hcluster_hg (id1, id2, weight):
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
4 """
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
5 import argparse
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
6 import math
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
7 from collections import OrderedDict
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
8
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
9
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
10 def main():
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
11 parser = argparse.ArgumentParser()
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
12
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
13 parser.add_argument('-i', metavar='in-file', type=argparse.FileType('rt'), required=True, help='Path to input file')
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
14
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
15 parser.add_argument('-o', metavar='out-file', type=argparse.FileType('wt'), required=True, help='Path to output file')
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
16
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
17 parser.add_argument('-r', action='store_true', default=False,
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
18 dest='reciprocal',
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
19 help='Annotate homolog pair')
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
20
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
21 parser.add_argument('--version', action='version', version='%(prog)s 1.0')
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
22
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
23 options = parser.parse_args()
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
24
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
25 results = OrderedDict()
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
26
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
27 for line in options.i:
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
28 line = line.rstrip()
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
29 line_cols = line.split('\t')
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
30 sequence1_id = line_cols[0]
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
31 sequence2_id = line_cols[1]
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
32 evalue = float(line_cols[10])
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
33
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
34 # Ignore self-matching hits
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
35 if sequence1_id != sequence2_id:
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
36 # Convert evalue to an integer weight with max 100
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
37 weight = 100
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
38
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
39 # If the evalue is 0, leave weight at 100
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
40 if evalue != 0.0:
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
41 weight = min(100, round(math.log10(evalue) / -2.0))
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
42
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
43 if (sequence1_id, sequence2_id) not in results:
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
44 results[(sequence1_id, sequence2_id)] = weight
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
45 else:
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
46 results[(sequence1_id, sequence2_id)] = max(results[(sequence1_id, sequence2_id)], weight)
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
47
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
48 for (sequence1_id, sequence2_id), weight in results.items():
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
49 if not options.reciprocal or (sequence2_id, sequence1_id) in results:
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
50 options.o.write("%s\t%s\t%d\n" % (sequence1_id, sequence2_id, weight))
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
51
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
52
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
53 if __name__ == "__main__":
70df762b48a8 planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/blast_parser commit 32272744ad83a704fd427b48aae574496a279901-dirty
earlhaminst
parents:
diff changeset
54 main()