Mercurial > repos > iuc > extract_metaphlan_database

annotate formatoutput.py @ 12:db58d5898fea draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 8ab597f47a3a12919b55055380b6e725879cc5c0
author iuc
date Sun, 08 Sep 2024 06:59:37 +0000
parents b4c750fe0fe3
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
1 #!/usr/bin/env python
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
2 # -*- coding: utf-8 -*-
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
3
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
4 import argparse
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
5 import re
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
6 from pathlib import Path
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
7
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
8 taxo_level = {
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
9 'k': 'kingdom',
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
10 'p': 'phylum',
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
11 'c': 'class',
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
12 'o': 'order',
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
13 'f': 'family',
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
14 'g': 'genus',
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
15 's': 'species',
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
16 't': 'strains'}
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
17
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
18
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
19 def split_levels(metaphlan_output_fp, out_dp, legacy_output):
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
20 '''
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
21 Split default MetaPhlAn into a report for each taxonomic level
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
22
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
23 :param metaphlan_output_fp: Path default MetaPhlAn output
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
24 :param out_dp: Path to output directory
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
25 :param legacy_output: Boolean for legacy output
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
26 '''
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
27 # prepare output files
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
28 abund_f = {
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
29 'k': open(out_dp / Path('kingdom'), 'w'),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
30 'p': open(out_dp / Path('phylum'), 'w'),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
31 'c': open(out_dp / Path('class'), 'w'),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
32 'o': open(out_dp / Path('order'), 'w'),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
33 'f': open(out_dp / Path('family'), 'w'),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
34 'g': open(out_dp / Path('genus'), 'w'),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
35 's': open(out_dp / Path('species'), 'w'),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
36 't': open(out_dp / Path('strains'), 'w')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
37 }
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
38 for level in abund_f:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
39 abund_f[level].write("%s\t" % taxo_level[level])
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
40 if not legacy_output:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
41 abund_f[level].write("%s_id\t" % taxo_level[level])
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
42 abund_f[level].write("abundance\n")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
43
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
44 levels_number = len(taxo_level)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
45
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
46 with open(metaphlan_output_fp, 'r') as metaphlan_output_f:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
47 with open(out_dp / Path('all'), 'w') as all_level_f:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
48 # write header in all leve file
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
49 for level in ['k', 'p', 'c', 'o', 'f', 'g', 's', 't']:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
50 all_level_f.write("%s\t" % taxo_level[level])
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
51 if not legacy_output:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
52 all_level_f.write("%s_id\t" % taxo_level[level])
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
53 all_level_f.write("abundance\n")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
54
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
55 # parse metaphlan file
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
56 for line in metaphlan_output_f.readlines():
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
57 # skip headers
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
58 if line.startswith("#"):
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
59 continue
5
b4c750fe0fe3 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit f3a89af137b13715f9fb13383577aceb2c445ce6
iuc
parents: 2
diff changeset
60
b4c750fe0fe3 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit f3a89af137b13715f9fb13383577aceb2c445ce6
iuc
parents: 2
diff changeset
61 # skip UNKNOWN (v3) or UNCLASSIFIED (v4) lines in predicted taxon relative abundances
b4c750fe0fe3 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit f3a89af137b13715f9fb13383577aceb2c445ce6
iuc
parents: 2
diff changeset
62 if "UNKNOWN" in line or 'UNCLASSIFIED' in line:
2
b6ecdfac241f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit f1c6f4fe1e572ace84cf9106bc253603f55aac55"
iuc
parents: 1
diff changeset
63 continue
5
b4c750fe0fe3 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit f3a89af137b13715f9fb13383577aceb2c445ce6
iuc
parents: 2
diff changeset
64
1
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
65 # spit lines
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
66 split_line = line[:-1].split('\t')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
67 taxo_n = split_line[0].split('|')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
68 if legacy_output:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
69 abundance = split_line[1]
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
70 else:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
71 taxo_id = split_line[1].split('|')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
72 abundance = split_line[2]
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
73
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
74 # get taxon name and ids
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
75 for i in range(len(taxo_n)):
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
76 taxo = taxo_n[i].split('__')[1]
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
77 taxo = taxo.replace("_", " ")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
78 all_level_f.write("%s\t" % taxo)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
79 if not legacy_output:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
80 all_level_f.write("%s\t" % taxo_id[i])
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
81
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
82 # if not all taxon levels
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
83 for i in range(len(taxo_n), levels_number):
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
84 all_level_f.write('\t')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
85
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
86 all_level_f.write("%s\n" % abundance)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
87
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
88 # write
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
89 last_taxo_level = taxo_n[-1].split('__')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
90 taxo = last_taxo_level[1].replace("_", " ")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
91 level = last_taxo_level[0]
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
92 abund_f[level].write("%s\t" % taxo)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
93 if not legacy_output:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
94 abund_f[level].write("%s\t" % taxo_id[-1])
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
95 abund_f[level].write("%s\n" % abundance)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
96
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
97 # close files
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
98 for taxo_level_f in abund_f:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
99 abund_f[taxo_level_f].close()
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
100
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
101
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
102 def format_for_krona(metaphlan_output_fp, krona_out_fp):
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
103 '''
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
104 Split default MetaPhlAn into a report for each taxonomic levKRONAel
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
105
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
106 :param metaphlan_output_fp: Path default MetaPhlAn output
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
107 :param krona_out: Path to output file for Krona
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
108 '''
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
109 re_replace = re.compile(r"\w__")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
110 re_bar = re.compile(r"\|")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
111 re_underscore = re.compile(r"_")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
112
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
113 with open(metaphlan_output_fp, 'r') as metaphlan_output_f:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
114 with open(krona_out_fp, 'w') as krona_out_f:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
115 for line in metaphlan_output_f.readlines():
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
116 if "s__" in line:
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
117 x = line.rstrip().split('\t')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
118 lineage = re.sub(re_bar, '', x[0])
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
119 lineage = re.sub(re_replace, '\t', lineage)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
120 lineage = re.sub(re_underscore, ' ', lineage)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
121 krona_out_f.write("%s\t%s\n" % (x[-1], lineage))
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
122
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
123
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
124 if __name__ == '__main__':
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
125 parser = argparse.ArgumentParser(description='Format MetaPhlAn output')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
126 subparsers = parser.add_subparsers(dest='function')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
127 # split_levels
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
128 split_levels_parser = subparsers.add_parser('split_levels', help='Split default MetaPhlAn into a report for each taxonomic level')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
129 split_levels_parser.add_argument('--metaphlan_output', help="Path to default MetaPhlAn output")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
130 split_levels_parser.add_argument('--outdir', help="Path to output directory")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
131 split_levels_parser.add_argument('--legacy-output', dest='legacy_output', action='store_true', help="Old MetaPhlAn2 two columns output")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
132 split_levels_parser.set_defaults(legacy_output=False)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
133 # format_for_krona
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
134 format_for_krona_parser = subparsers.add_parser('format_for_krona', help='Split default MetaPhlAn into a report for each taxonomic level')
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
135 format_for_krona_parser.add_argument('--metaphlan_output', help="Path to default MetaPhlAn output")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
136 format_for_krona_parser.add_argument('--krona_output', help="Path to Krona output directory")
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
137
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
138 args = parser.parse_args()
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
139
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
140 if args.function == 'split_levels':
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
141 split_levels(
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
142 Path(args.metaphlan_output),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
143 Path(args.outdir),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
144 args.legacy_output)
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
145 elif args.function == 'format_for_krona':
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
146 format_for_krona(
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
147 Path(args.metaphlan_output),
1aaa9b943a83 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/metaphlan/ commit 2b87bc7417360e2b2c9ec0605d475909f6f0482f"
iuc
parents:
diff changeset
148 Path(args.krona_output))