Mercurial > repos > public-health-bioinformatics > screen_abricate_report

annotate screen_abricate_report.py @ 2:912a3a3dc082 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/screen_abricate_report commit 2ec76aac2fcf466fc16091bfff8b7cb83fd92467"
author public-health-bioinformatics
date Thu, 19 Dec 2019 20:31:11 -0500
parents 4f963b583186
children 2262e531c50b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
1 #!/usr/bin/env python
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
2
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
3 from __future__ import print_function
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
4
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
5 import argparse
1
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
6 import csv
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
7 import re
1
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
8
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
9
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
10 def parse_screen_file(screen_file):
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
11 screen = []
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
12 with open(screen_file) as f:
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
13 reader = csv.DictReader(f, delimiter="\t", quotechar='"')
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
14 for row in reader:
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
15 screen.append(row)
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
16 return screen
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
17
1
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
18
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
19 def get_fieldnames(input_file):
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
20 with open(input_file) as f:
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
21 reader = csv.DictReader(f, delimiter="\t", quotechar='"')
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
22 row = next(reader)
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
23 fieldnames = row.keys()
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
24 return fieldnames
1
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
25
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
26
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
27 def main(args):
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
28 screen = parse_screen_file(args.screening_file)
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
29 gene_detection_status_fieldnames = ['gene_name', 'detected']
1
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
30 with open(args.abricate_report, 'r') as f1, \
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
31 open(args.screened_report, 'w') as f2, \
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
32 open(args.gene_detection_status, 'w') as f3:
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
33 abricate_report_reader = csv.DictReader(f1, delimiter="\t", quotechar='"')
1
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
34 screened_report_writer = csv.DictWriter(f2, delimiter="\t", quotechar='"',
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
35 fieldnames=abricate_report_reader.fieldnames)
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
36 gene_detection_status_writer = csv.DictWriter(f3, delimiter="\t", quotechar='"',
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
37 fieldnames=gene_detection_status_fieldnames)
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
38 screened_report_writer.writeheader()
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
39 gene_detection_status_writer.writeheader()
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
40
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
41 for gene in screen:
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
42 gene_detection_status = {
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
43 'gene_name': gene['gene_name'],
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
44 'detected': False
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
45 }
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
46 for abricate_report_row in abricate_report_reader:
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
47 if re.search(gene['regex'], abricate_report_row['GENE']):
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
48 gene_detection_status['detected'] = True
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
49 screened_report_writer.writerow(abricate_report_row)
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
50 gene_detection_status_writer.writerow(gene_detection_status)
1
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
51 f1.seek(0) # return file pointer to start of abricate report
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
52
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
53
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
54 if __name__ == '__main__':
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
55 parser = argparse.ArgumentParser()
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
56 parser.add_argument("abricate_report", help="Input: Abricate report to screen (tsv)")
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
57 parser.add_argument("--screening_file", help="Input: List of genes to screen for (tsv)")
1
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
58 parser.add_argument("--screened_report", help=("Output: Screened abricate report ",
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
59 "including only genes of interest (tsv)"))
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
60 parser.add_argument("--gene_detection_status", help=("Output: detection status for all genes ",
4f963b583186 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit c27095ceaf23aa2c910e0ea866db9ec8d3e7816b"
public-health-bioinformatics
parents: 0
diff changeset
61 "listed in the screening file (tsv)"))
0
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
62 args = parser.parse_args()
b2d56a44a872 "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/screen_abricate_report commit 1d569fc27b052d1982d82ca19455caaff6386f46"
public-health-bioinformatics
parents:
diff changeset
63 main(args)