Mercurial > repos > dfornika > blast_report_basic

annotate blast_report.py @ 0:5dfd84907521 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
author dfornika
date Tue, 03 Mar 2020 00:14:34 +0000
parents
children 386a88793078
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
1 #!/usr/bin/env python
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
2 '''Report on BLAST results.
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
3
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
4 python blast_report.py input_tab cheetah_tmpl output_html output_tab [-f [filter_pident]:[filterkw1,...,filterkwN]] [-b bin1_label=bin1_path[,...binN_label=binN_path]]
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
5 '''
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
6 import argparse
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
7 import re
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
8 import sys
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
9
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
10 from Cheetah.Template import Template
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
11
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
12
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
13 def stop_err( msg ):
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
14 sys.stderr.write("%s\n" % msg)
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
15 sys.exit(1)
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
16
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
17
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
18 class BLASTBin:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
19 def __init__(self, label, file):
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
20 self.label = label
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
21 self.dict = {}
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
22
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
23 file_in = open(file)
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
24 for line in file_in:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
25 self.dict[line.rstrip().split('.')[0]] = ''
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
26 file_in.close()
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
27
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
28 def __str__(self):
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
29 return "label: %s dict: %s" % (self.label, str(self.dict))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
30
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
31
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
32 class BLASTQuery:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
33 def __init__(self, query_id):
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
34 self.query_id = query_id
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
35 self.matches = []
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
36 self.match_accessions = {}
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
37 self.bins = {} #{bin(label):[match indexes]}
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
38 self.pident_filtered = 0
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
39 self.kw_filtered = 0
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
40 self.kw_filtered_breakdown = {} #{kw:count}
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
41
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
42 def __str__(self):
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
43 return "query_id: %s len(matches): %s bins (labels only): %s pident_filtered: %s kw_filtered: %s kw_filtered_breakdown: %s" \
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
44 % (self.query_id,
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
45 str(len(self.matches)),
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
46 str([bin.label for bin in bins]),
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
47 str(self.pident_filtered),
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
48 str(self.kw_filtered),
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
49 str(self.kw_filtered_breakdown))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
50
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
51
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
52 class BLASTMatch:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
53 def __init__(self, subject_acc, subject_descr, score, p_cov, p_ident, subject_bins):
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
54 self.subject_acc = subject_acc
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
55 self.subject_descr = subject_descr
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
56 self.score = score
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
57 self.p_cov = p_cov
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
58 self.p_ident = p_ident
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
59 self.bins = subject_bins
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
60
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
61 def __str__(self):
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
62 return "subject_acc: %s subject_descr: %s score: %s p-cov: %s p-ident: %s" \
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
63 % (self.subject_acc,
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
64 self.subject_descr,
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
65 str(self.score),
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
66 str(round(self.p_cov,2)),
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
67 str(round(self.p_ident, 2)))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
68
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
69
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
70
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
71 #PARSE OPTIONS AND ARGUMENTS
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
72 parser = argparse.ArgumentParser()
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
73
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
74 parser.add_argument('-f', '--filter',
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
75 type='string',
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
76 dest='filter',
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
77 )
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
78 parser.add_argument('-b', '--bins',
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
79 type='string',
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
80 dest='bins'
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
81 )
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
82 parser.add_argument('-r', '--redundant',
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
83 dest='redundant',
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
84 default=False,
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
85 action='store_true'
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
86 )
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
87 args = parser.parse_args()
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
88
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
89 try:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
90 input_tab, cheetah_tmpl, output_html, output_tab = args
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
91 except:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
92 stop_err('you must supply the arguments input_tab, cheetah_tmpl and output_html.')
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
93 # print('input_tab: %s cheetah_tmpl: %s output_html: %s output_tab: %s' % (input_tab, cheetah_tmpl, output_html, output_tab))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
94
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
95
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
96 #BINS
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
97 bins=[]
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
98 if args.bins != None:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
99 bins = list([BLASTBin(label_file.split('=')[0],label_file.split('=')[-1]) for label_file in args.bins.split(',')])
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
100 print('database bins: %s' % str([bin.label for bin in bins]))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
101
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
102 #FILTERS
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
103 filter_pident = 0
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
104 filter_kws = []
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
105 if args.filter != None:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
106 pident_kws = args.filter.split(':')
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
107 filter_pident = float(pident_kws[0])
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
108 filter_kws = pident_kws[-1].split(',')
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
109 print('filter_pident: %s filter_kws: %s' % (str(filter_pident), str(filter_kws)))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
110
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
111 if args.redundant:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
112 print('Throwing out redundant hits...')
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
113
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
114 #RESULTS!
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
115 PIDENT_COL = 2
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
116 DESCR_COL = 25
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
117 SUBJ_ID_COL = 12
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
118 SCORE_COL = 11
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
119 PCOV_COL = 24
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
120 queries = []
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
121 current_query = ''
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
122 output_tab = open(output_tab, 'w')
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
123
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
124 with open(input_tab) as input_tab:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
125 for line in input_tab:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
126 cols = line.split('\t')
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
127 if cols[0] != current_query:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
128 current_query = cols[0]
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
129 queries.append(BLASTQuery(current_query))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
130
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
131 try:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
132 accs = cols[SUBJ_ID_COL].split('|')[1::2][1::2]
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
133 except IndexError as e:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
134 stop_err("Problem with splitting:" + cols[SUBJ_ID_COL])
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
135
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
136 #hsp option: keep best (first) hit only for each query and accession id.
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
137 if args.redundant:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
138 if accs[0] in queries[-1].match_accessions:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
139 continue #don't save the result and skip to the next
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
140 else:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
141 queries[-1].match_accessions[accs[0]] = ''
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
142
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
143
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
144 p_ident = float(cols[PIDENT_COL])
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
145 #FILTER BY PIDENT
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
146 if p_ident < filter_pident: #if we are not filtering, filter_pident == 0 and this will never evaluate to True
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
147 queries[-1].pident_filtered += 1
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
148 continue
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
149
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
150 descrs = cols[DESCR_COL]
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
151 #FILTER BY KEY WORDS
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
152 filter_by_kw = False
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
153 for kw in filter_kws:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
154 kw = kw.strip() #Fix by Damion D Nov 2013
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
155 if kw != '' and re.search(kw, descrs, re.IGNORECASE):
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
156 filter_by_kw = True
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
157 try:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
158 queries[-1].kw_filtered_breakdown[kw] += 1
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
159 except:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
160 queries[-1].kw_filtered_breakdown[kw] = 1
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
161 if filter_by_kw: #if we are not filtering, for loop will not be entered and this will never be True
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
162 queries[-1].kw_filtered += 1
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
163 continue
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
164 descr = descrs.split(';')[0]
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
165
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
166 #ATTEMPT BIN
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
167 subj_bins = []
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
168 for bin in bins: #if we are not binning, bins = [] so for loop not entered
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
169 for acc in accs:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
170 if acc.split('.')[0] in bin.dict:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
171 try:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
172 queries[-1].bins[bin.label].append(len(queries[-1].matches))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
173 except:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
174 queries[-1].bins[bin.label] = [len(queries[-1].matches)]
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
175 subj_bins.append(bin.label)
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
176 break #this result has been binned to this bin so break
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
177 acc = accs[0]
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
178
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
179 score = int(float(cols[SCORE_COL]))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
180 p_cov = float(cols[PCOV_COL])
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
181
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
182 #SAVE RESULT
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
183 queries[-1].matches.append(
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
184 BLASTMatch(acc, descr, score, p_cov, p_ident, subj_bins)
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
185 )
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
186 output_tab.write(line)
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
187 input_tab.close()
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
188 output_tab.close()
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
189
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
190 '''
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
191 for query in queries:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
192 print(query)
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
193 for match in query.matches:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
194 print(' %s' % str(match))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
195 for bin in query.bins:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
196 print(' bin: %s' % bin)
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
197 for x in query.bins[bin]:
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
198 print(' %s' % str(query.matches[x]))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
199 '''
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
200
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
201 namespace = {'queries': queries}
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
202 html = Template(file=cheetah_tmpl, searchList=[namespace])
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
203 out_html = open(output_html, 'w')
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
204 out_html.write(str(html))
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
205 out_html.close()
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
206
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
207
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
208 if __name__ == '__main__':
5dfd84907521 planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/blast_report_basic commit bc359460bb66db7946cc68ccbd47cd479624c4a1-dirty
dfornika
parents:
diff changeset
209 main()