annotate amino2consensus.py @ 29:4ff24c044fed draft default tip

"planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
author dfornika
date Thu, 27 Feb 2020 22:31:32 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
29
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
1 #!/usr/bin/env python
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
2
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
3 from __future__ import print_function
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
4
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
5 import argparse
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
6 import csv
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
7
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
8
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
9 AMINO_ACIDS = ['A', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'K', 'L', 'M', 'N', 'P', 'Q', 'R', 'S', 'T', 'V', 'W', 'Y', '*']
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
10
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
11
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
12 def determine_amino(amino_counts, threshold):
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
13 amino = ""
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
14 total_count = sum(amino_counts.values())
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
15 amino_with_max_counts = sorted(amino_counts, key=amino_counts.get, reverse=True)[0]
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
16 if total_count == 0:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
17 amino = "#"
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
18 elif (amino_counts[amino_with_max_counts] / float(total_count)) > threshold:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
19 amino = amino_with_max_counts
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
20 else:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
21 amino = "@"
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
22 return amino
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
23
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
24
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
25 def determine_first_region(amino_file):
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
26 with open(amino_file) as f:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
27 reader = csv.DictReader(f)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
28 row = next(reader)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
29 region = row['region']
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
30 return region
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
31
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
32
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
33 def main(args):
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
34 current_region = determine_first_region(args.amino)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
35 seq = []
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
36 with open(args.amino) as f:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
37 reader = csv.DictReader(f)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
38 for row in reader:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
39 if row['region'] == current_region:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
40 amino_counts = {}
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
41 for amino_acid in AMINO_ACIDS:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
42 amino_counts[amino_acid] = int(row[amino_acid])
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
43 amino = determine_amino(amino_counts, args.threshold)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
44 seq.append(amino)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
45 else:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
46 print(">" + current_region)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
47 print(''.join(seq))
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
48 current_region = row['region']
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
49 seq = []
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
50 amino_counts = {}
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
51 for amino_acid in AMINO_ACIDS:
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
52 amino_counts[amino_acid] = int(row[amino_acid])
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
53 amino = determine_amino(amino_counts, args.threshold)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
54 seq.append(amino)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
55 print(">" + current_region)
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
56 print(''.join(seq))
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
57
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
58
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
59 if __name__ == '__main__':
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
60 parser = argparse.ArgumentParser()
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
61 parser.add_argument("amino", help="MiCall amino.csv output file")
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
62 parser.add_argument("--threshold", default=0.15, type=float, help="Threshold for calling")
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
63 args = parser.parse_args()
4ff24c044fed "planemo upload for repository https://github.com/public-health-bioinformatics/galaxy_tools/blob/master/tools/micall-lite commit 9d563c366233a5de79429ac1fa8f994f5d8f785d-dirty"
dfornika
parents:
diff changeset
64 main(args)