Mercurial > repos > iuc > humann_barplot
annotate customizemapping.py @ 2:3f16956afb8a draft
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 9d1eca3087b58ff03fff0cd42a3092e2b4d86f60
author | iuc |
---|---|
date | Thu, 18 Aug 2022 14:32:57 +0000 |
parents | 5240d62d864d |
children |
rev | line source |
---|---|
0
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
1 #!/usr/bin/env python |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
2 # -*- coding: utf-8 -*- |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
3 |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
4 import argparse |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
5 from pathlib import Path |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
6 |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
7 |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
8 if __name__ == '__main__': |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
9 # Read command line |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
10 parser = argparse.ArgumentParser(description='Customize HUMAnN utility mapping') |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
11 parser.add_argument('--in_mapping', help="Path to mapping file to reduce") |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
12 parser.add_argument('--features', help="Path to tabular file with features to keep in first column") |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
13 parser.add_argument('--elements', help="Path to tabular file with elements to keep in other columns") |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
14 parser.add_argument('--out_mapping', help="Path to reduced mapping file") |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
15 args = parser.parse_args() |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
16 |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
17 in_mapping_fp = Path(args.in_mapping) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
18 feature_fp = Path(args.features) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
19 element_fp = Path(args.elements) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
20 out_mapping_fp = Path(args.out_mapping) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
21 |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
22 # extract features to keep |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
23 features = set() |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
24 with open(feature_fp, 'r') as feature_f: |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
25 for line in feature_f.readlines(): |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
26 features.add(line.split("\t")[0]) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
27 print(features) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
28 |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
29 # extract elements to keep |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
30 elements = set() |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
31 with open(element_fp, 'r') as element_f: |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
32 for line in element_f.readlines(): |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
33 elements.add(line.split("\t")[0]) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
34 print(elements) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
35 |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
36 # write mapping for features to keep while keeping only elements |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
37 with open(in_mapping_fp, 'r') as in_mapping_f: |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
38 with open(out_mapping_fp, 'w') as out_mapping_f: |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
39 for line in in_mapping_f.readlines(): |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
40 l_split = line.split("\t") |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
41 feat = l_split[0] |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
42 if feat in features: |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
43 to_write = [feat] |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
44 for e in l_split[1:]: |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
45 if e in elements: |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
46 to_write.append(e) |
5240d62d864d
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff
changeset
|
47 out_mapping_f.write("%s\n" % '\t'.join(to_write)) |