Mercurial > repos > iuc > humann_split_table

annotate customizemetadata.py @ 3:cc6a10100a14 draft

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/humann commit 32d4c566afe55179531fdd106dc5a996cc09bbb1
author iuc
date Wed, 29 Mar 2023 20:01:50 +0000
parents 44cacbe09bfc
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
506d4a349aa1 "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
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
2 # -*- coding: utf-8 -*-
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
3
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
4 import argparse
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
5 import bz2
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
6 import json
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
7 import pickle
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
8 import re
2
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
9 import sys
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
10 from importlib.metadata import version
0
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
11 from pathlib import Path
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
12
2
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
13 from packaging.version import Version
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
14
0
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
15
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
16 def load_from_json(json_fp):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
17 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
18 Read JSON file with marker metadata
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
19
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
20 :param json_fp: Path to JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
21 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
22 with open(json_fp, 'r') as json_f:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
23 data = json.load(json_f)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
24
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
25 for m in data['markers']:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
26 data['markers'][m]['ext'] = set(data['markers'][m]['ext'])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
27
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
28 for t in data['taxonomy']:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
29 if isinstance(data['taxonomy'][t], list):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
30 data['taxonomy'][t] = tuple(data['taxonomy'][t])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
31 return data
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
32
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
33
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
34 def dump_to_json(data, json_fp):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
35 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
36 Dump marker metadata to JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
37
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
38 :param json_fp: Path to JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
39 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
40 for m in data['markers']:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
41 data['markers'][m]['ext'] = list(data['markers'][m]['ext'])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
42
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
43 with open(json_fp, 'w') as json_f:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
44 json.dump(data, json_f)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
45
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
46
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
47 def transform_pkl_to_json(pkl_fp, json_fp):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
48 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
49 Read Pickle file and drop it to a JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
50
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
51 :param pkl_fp: Path to input Pickle file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
52 :param json_fp: Path to output JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
53 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
54 # load metadata from Pickle file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
55 with bz2.BZ2File(pkl_fp, 'r') as pkl_f:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
56 in_metadata = pickle.load(pkl_f)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
57
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
58 out_metadata = {
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
59 'markers': in_metadata['markers'],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
60 'taxonomy': in_metadata['taxonomy'],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
61 'merged_taxon': {}
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
62 }
2
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
63
0
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
64 # transform merged_taxons tuple keys to string
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
65 for k in in_metadata['merged_taxon']:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
66 n = ' , '.join(k)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
67 out_metadata[n] = in_metadata['merged_taxon'][k]
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
68
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
69 # dump metadata to JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
70 dump_to_json(out_metadata, json_fp)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
71
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
72
2
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
73 def validate_map_version(infile, file_type):
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
74 '''
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
75 Check conformity of a user-provided pkl file to Metaphlan SGB (>= v4.0).
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
76
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
77 :param infile: Path to input Pickle/JSON file
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
78 :param file_type: String definining file type, pkl or JSON. Case-insensitive
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
79 '''
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
80 file_type = file_type.lower()
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
81 if file_type == 'pkl' or file_type == 'pickle':
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
82 # load metadata from Pickle file
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
83 with bz2.BZ2File(infile, 'r') as pkl_f:
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
84 in_metadata = pickle.load(pkl_f)
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
85 elif file_type == 'json':
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
86 in_metadata = load_from_json(infile)
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
87 else:
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
88 raise ValueError("Unsupported file type to validate.")
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
89
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
90 # Get metaphlan version in $PATH
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
91 metaphlan_version = Version(version('metaphlan'))
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
92
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
93 # Ensure that there are 8 taxonomy levels separated with "|"s.
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
94 # v3 DB release encodes the taxids as: ('2|1224|1236|91347|543|547|354276', 4404432)
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
95 # v4 DB release encodes the taxids as: ('2|1224|1236|91347|543|547|354276|', 4404432)
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
96 for k in in_metadata['taxonomy']:
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
97 if (in_metadata['taxonomy'][k][0].count('|') != 7 and metaphlan_version >= Version('4')) or (in_metadata['taxonomy'][k][0].count('|') != 6 and metaphlan_version < Version('4')):
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
98 # raise ValueError("Missing/Extra values in GCA list")
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
99 print("The input taxonomy mapping file %s is incompatible with Metaphlan v.%s in $PATH." % (infile, metaphlan_version))
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
100 sys.exit(42)
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
101
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
102 print("%s is compatible with Metaphlan v.%s." % (infile, metaphlan_version))
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
103
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
104
0
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
105 def transform_json_to_pkl(json_fp, pkl_fp):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
106 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
107 Read JSON file and drop it to a Pickle file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
108
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
109 :param json_fp: Path to input JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
110 :param pkl_fp: Path to output Pickle file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
111 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
112 # load metadata from JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
113 in_metadata = load_from_json(json_fp)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
114
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
115 out_metadata = {
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
116 'markers': in_metadata['markers'],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
117 'taxonomy': in_metadata['taxonomy'],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
118 'merged_taxon': {}
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
119 }
2
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
120
0
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
121 # transform merged_taxons keys to tuple
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
122 for k in in_metadata['merged_taxon']:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
123 n = ' , '.split(k)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
124 out_metadata[n] = in_metadata['merged_taxon'][k]
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
125
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
126 # dump metadata to Pickle file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
127 with bz2.BZ2File(pkl_fp, 'w') as pkl_f:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
128 pickle.dump(out_metadata, pkl_f)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
129
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
130
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
131 def add_marker(in_json_fp, out_json_fp, name, m_length, g_length, gca, k_name, k_id, p_name, p_id, c_name, c_id, o_name, o_id, f_name, f_id, g_name, g_id, s_name, s_id, t_name):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
132 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
133 Add marker to JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
134
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
135 :param in_json_fp: Path to input JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
136 :param out_json_fp: Path to output JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
137 :param name: Name of new marker
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
138 :param m_length: Length of new marker
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
139 :param g_length: List with lengths of genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
140 :param gca: List with GCA of genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
141 :param k_name: List with Name of Kingdom for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
142 :param k_id: List with NCBI id of Kingdom for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
143 :param p_name: List with Name of Phylum for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
144 :param p_id: List with NCBI id of Phylum for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
145 :param c_name: List with Name of Class for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
146 :param c_id: List with NCBI id of Class for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
147 :param o_name: List with Name of Order for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
148 :param o_id: List with NCBI id of Order for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
149 :param f_name: List with Name of Family for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
150 :param f_id: List with NCBI id of Family for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
151 :param g_name: List with Name of Genus for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
152 :param g_id: List with NCBI id of Genus for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
153 :param s_name: List with Name of Species for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
154 :param s_id: List with NCBI id of Species for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
155 :param t_name: List with Name of Strain for genomes from which the new marker has been extracted
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
156 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
157 metadata = load_from_json(in_json_fp)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
158
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
159 # check that all lists have same size
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
160 genome_n = len(g_length)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
161 if len(gca) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
162 raise ValueError("Missing/Extra values in GCA list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
163 if len(k_name) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
164 raise ValueError("Missing/Extra values in Kingdom name list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
165 if len(k_id) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
166 raise ValueError("Missing/Extra values in Kingdom ID list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
167 if len(p_name) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
168 raise ValueError("Missing/Extra values in Phylum name list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
169 if len(p_id) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
170 raise ValueError("Missing/Extra values in Phylum ID list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
171 if len(c_name) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
172 raise ValueError("Missing/Extra values in Class name list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
173 if len(c_id) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
174 raise ValueError("Missing/Extra values in Class ID list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
175 if len(o_name) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
176 raise ValueError("Missing/Extra values in Order name list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
177 if len(o_id) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
178 raise ValueError("Missing/Extra values in Order ID list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
179 if len(f_name) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
180 raise ValueError("Missing/Extra values in Family name list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
181 if len(f_id) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
182 raise ValueError("Missing/Extra values in Family ID list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
183 if len(g_name) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
184 raise ValueError("Missing/Extra values in Genus name list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
185 if len(g_id) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
186 raise ValueError("Missing/Extra values in Genus ID list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
187 if len(s_name) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
188 raise ValueError("Missing/Extra values in Species name list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
189 if len(s_id) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
190 raise ValueError("Missing/Extra values in Species ID list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
191 if len(t_name) != genome_n:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
192 raise ValueError("Missing/Extra values in Strain name list")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
193
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
194 # create dictionary to aggregate genome taxonomies and identify marker taxonomy
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
195 taxonomy = {
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
196 'k': set(),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
197 'p': set(),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
198 'c': set(),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
199 'o': set(),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
200 'f': set(),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
201 'g': set(),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
202 's': set(),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
203 't': set(),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
204 }
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
205
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
206 # parse genomes
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
207 for i in range(genome_n):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
208 # add taxonomy of new genome
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
209 g_taxo_names = "k__%s|p__%s|c__%s|o__%s|f__%s|g__%s|s__%s|t__%s" % (
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
210 k_name[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
211 p_name[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
212 c_name[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
213 o_name[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
214 f_name[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
215 g_name[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
216 s_name[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
217 t_name[i]
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
218 )
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
219 g_taxo_ids = "%s|%s|%s|%s|%s|%s|%s" % (
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
220 k_id[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
221 p_id[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
222 c_id[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
223 o_id[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
224 f_id[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
225 g_id[i],
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
226 s_id[i]
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
227 )
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
228 metadata['taxonomy'][g_taxo_names] = (g_taxo_ids, g_length[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
229 # aggregate taxon levels using sets
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
230 taxonomy['k'].add(k_name[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
231 taxonomy['p'].add(p_name[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
232 taxonomy['c'].add(c_name[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
233 taxonomy['o'].add(o_name[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
234 taxonomy['f'].add(f_name[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
235 taxonomy['g'].add(g_name[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
236 taxonomy['s'].add(s_name[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
237 taxonomy['t'].add(t_name[i])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
238
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
239 # extract clade and taxon of marker
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
240 clade = '' # last level before taxomy of genomes diverge
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
241 taxon = '' # combination of levels before divergence
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
242 for level in ['k', 'p', 'c', 'o', 'f', 'g', 's', 't']:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
243 taxo = list(taxonomy[level])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
244 if len(taxo) == 1:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
245 clade = taxo[0]
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
246 taxon = "%s|%s__%s" % (taxon, level, taxo)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
247
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
248 # add information about the new marker
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
249 metadata['markers'][name] = {
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
250 'clade': clade,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
251 'ext': set(gca),
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
252 'len': m_length,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
253 'taxon': taxon
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
254 }
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
255
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
256 dump_to_json(metadata, out_json_fp)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
257
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
258
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
259 def format_markers(marker_l):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
260 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
261 Format markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
262
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
263 :param marker_l: list of markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
264 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
265 markers = []
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
266 for m in marker_l:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
267 m = m.rstrip()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
268 if ' ' in m:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
269 markers.append(m.split(' ')[0])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
270 else:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
271 markers.append(m)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
272 return markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
273
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
274
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
275 def get_markers(marker_fp):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
276 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
277 Get markers from a file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
278
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
279 :param marker_fp: Path to file with markers (1 per line)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
280 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
281 # load markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
282 with open(marker_fp, 'r') as marker_f:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
283 markers = marker_f.readlines()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
284
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
285 # format markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
286 markers = format_markers(markers)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
287
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
288 return markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
289
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
290
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
291 def check_not_found_markers(found_markers, original_markers):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
292 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
293 Check list of markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
294
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
295 :param found_markers: list of found markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
296 :param original_markers: list of original markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
297 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
298 if len(found_markers) != len(original_markers):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
299 print('markers not found:')
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
300 for m in original_markers:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
301 if m not in found_markers:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
302 print('- "%s"' % m)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
303
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
304
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
305 def prune_taxonomy(in_taxonomy, taxon_s, gca_s):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
306 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
307 Prune taxonomy to keep only listed taxonomy
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
308
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
309 :param in_taxonomy: dictionary with list of taxonomy
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
310 :param taxon_s: set of taxons to keep
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
311 :param gca_s: set of GCA ids to keep
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
312 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
313 out_taxonomy = {}
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
314 kept_taxonomy = set()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
315 kept_taxons = set()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
316 kept_gca = set()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
317 for t, v in in_taxonomy.items():
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
318 # check if t match element in list of taxon_s
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
319 kept_taxon = False
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
320 for t_k in taxon_s:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
321 if t_k in t:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
322 kept_taxon = True
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
323 out_taxonomy[t] = v
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
324 kept_taxonomy.add(t)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
325 kept_taxons.add(t_k)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
326 break
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
327 # check if GCA in the taxon id
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
328 s = re.search(r'GCA_\d+$', t)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
329 if s:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
330 gca = s[0]
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
331 # check if GCA in taxon id is in the list GCA to keep
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
332 if gca in gca_s:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
333 kept_gca.add(gca)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
334 if not kept_taxon:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
335 out_taxonomy[t] = v
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
336 kept_taxonomy.add(t)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
337
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
338 print('%s kept taxonomy' % len(kept_taxonomy))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
339 print('%s / %s taxons not found' % (len(taxon_s) - len(kept_taxons), len(taxon_s)))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
340 print('%s / %s GCA taxons not found' % (len(gca_s) - len(kept_gca), len(gca_s)))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
341 return out_taxonomy
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
342
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
343
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
344 def remove_markers(in_json_fp, marker_fp, out_json_fp, kept_marker_fp):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
345 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
346 Remove markers from JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
347
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
348 :param in_json_fp: Path to input JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
349 :param marker_fp: Path to file with markers to remove (1 per line)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
350 :param out_json_fp: Path to output JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
351 :param kept_marker_fp: Path to file with kept markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
352 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
353 in_metadata = load_from_json(in_json_fp)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
354
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
355 # load markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
356 markers_to_remove = set(get_markers(marker_fp))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
357 print('%s markers to remove' % len(markers_to_remove))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
358
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
359 # keep merged_taxon
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
360 out_metadata = {
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
361 'markers': {},
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
362 'taxonomy': {},
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
363 'merged_taxon': in_metadata['merged_taxon']
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
364 }
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
365
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
366 # parse markers to keep
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
367 removed_markers = []
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
368 kept_markers = []
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
369 taxons_to_keep = set()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
370 gca_to_keep = set()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
371 for m, v in in_metadata['markers'].items():
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
372 if m not in markers_to_remove:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
373 out_metadata['markers'][m] = v
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
374 kept_markers.append(m)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
375 taxons_to_keep.add(v['taxon'])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
376 gca_to_keep.update(v['ext'])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
377 else:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
378 removed_markers.append(m)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
379 print('%s removed markers' % len(removed_markers))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
380
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
381 # check markers that are not found
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
382 check_not_found_markers(removed_markers, markers_to_remove)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
383
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
384 # keep only taxonomy in taxons_to_keep or with GCA in gca_to_keep
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
385 out_metadata['taxonomy'] = prune_taxonomy(in_metadata['taxonomy'], taxons_to_keep, gca_to_keep)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
386
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
387 # save to JSON
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
388 dump_to_json(out_metadata, out_json_fp)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
389
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
390 # write list of kept markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
391 with open(kept_marker_fp, 'w') as kept_marker_f:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
392 for m in kept_markers:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
393 kept_marker_f.write("%s\n" % m)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
394
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
395
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
396 def keep_markers(in_json_fp, marker_fp, out_json_fp):
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
397 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
398 Keep markers from JSON file, others will be removed
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
399
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
400 :param in_json_fp: Path to input JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
401 :param marker_fp: Path to file with markers to keep (1 per line)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
402 :param out_json_fp: Path to output JSON file
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
403 '''
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
404 in_metadata = load_from_json(in_json_fp)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
405
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
406 # load markers
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
407 markers_to_keep = set(get_markers(marker_fp))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
408 print('%s markers to keep' % len(markers_to_keep))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
409
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
410 # keep merged_taxon
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
411 out_metadata = {
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
412 'markers': {},
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
413 'taxonomy': {},
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
414 'merged_taxon': in_metadata['merged_taxon']
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
415 }
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
416
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
417 # parse markers to keep
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
418 kept_markers = []
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
419 taxons_to_keep = set()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
420 gca_to_keep = set()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
421 for m, v in in_metadata['markers'].items():
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
422 if m in markers_to_keep:
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
423 out_metadata['markers'][m] = v
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
424 kept_markers.append(m)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
425 taxons_to_keep.add(v['taxon'])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
426 gca_to_keep.update(v['ext'])
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
427 print('%s kept markers' % len(kept_markers))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
428
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
429 # check markers that are not found
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
430 check_not_found_markers(kept_markers, markers_to_keep)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
431
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
432 # keep only taxonomy in taxons_to_keep or with GCA in gca_to_keep
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
433 out_metadata['taxonomy'] = prune_taxonomy(in_metadata['taxonomy'], taxons_to_keep, gca_to_keep)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
434
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
435 # save to JSON
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
436 dump_to_json(out_metadata, out_json_fp)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
437
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
438
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
439 if __name__ == '__main__':
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
440 # Read command line
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
441 parser = argparse.ArgumentParser(description='Customize MetaPhlan database')
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
442 subparsers = parser.add_subparsers(dest='function')
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
443 # transform_pkl_to_json subcommand
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
444 pkl_to_json_parser = subparsers.add_parser('transform_pkl_to_json', help='Transform Pickle to JSON to get marker metadata')
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
445 pkl_to_json_parser.add_argument('--pkl', help="Path to input Pickle file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
446 pkl_to_json_parser.add_argument('--json', help="Path to output JSON file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
447 # transform_json_to_pkl subcommand
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
448 json_to_pkl_parser = subparsers.add_parser('transform_json_to_pkl', help='Transform JSON to Pickle to push marker metadata')
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
449 json_to_pkl_parser.add_argument('--json', help="Path to input JSON file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
450 json_to_pkl_parser.add_argument('--pkl', help="Path to output Pickle file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
451 # add_marker subcommand
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
452 add_marker_parser = subparsers.add_parser('add_marker', help='Add new marker to JSON file')
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
453 add_marker_parser.add_argument('--in_json', help="Path to input JSON file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
454 add_marker_parser.add_argument('--out_json', help="Path to output JSON file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
455 add_marker_parser.add_argument('--name', help="Name of new marker")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
456 add_marker_parser.add_argument('--m_length', help="Length of new marker")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
457 add_marker_parser.add_argument('--g_length', help="Length of genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
458 add_marker_parser.add_argument('--gca', help="GCA of genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
459 add_marker_parser.add_argument('--k_name', help="Name of Kingdom for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
460 add_marker_parser.add_argument('--k_id', help="NCBI id of Kingdom for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
461 add_marker_parser.add_argument('--p_name', help="Name of Phylum for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
462 add_marker_parser.add_argument('--p_id', help="NCBI id of Phylum for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
463 add_marker_parser.add_argument('--c_name', help="Name of Class for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
464 add_marker_parser.add_argument('--c_id', help="NCBI id of Class for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
465 add_marker_parser.add_argument('--o_name', help="Name of Order for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
466 add_marker_parser.add_argument('--o_id', help="NCBI id of Order for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
467 add_marker_parser.add_argument('--f_name', help="Name of Family for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
468 add_marker_parser.add_argument('--f_id', help="NCBI id of Family for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
469 add_marker_parser.add_argument('--g_name', help="Name of Genus for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
470 add_marker_parser.add_argument('--g_id', help="NCBI id of Genus for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
471 add_marker_parser.add_argument('--s_name', help="Name of Species for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
472 add_marker_parser.add_argument('--s_id', help="NCBI id of Species for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
473 add_marker_parser.add_argument('--t_name', help="Name of Strain for genome from which the new marker has been extracted", action="append")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
474 # remove_markers subcommand
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
475 remove_markers_parser = subparsers.add_parser('remove_markers', help='Remove markers from JSON file')
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
476 remove_markers_parser.add_argument('--in_json', help="Path to input JSON file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
477 remove_markers_parser.add_argument('--markers', help="Path to file with markers to remove (1 per line)")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
478 remove_markers_parser.add_argument('--out_json', help="Path to output JSON file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
479 remove_markers_parser.add_argument('--kept_markers', help="Path to file with kept markers")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
480 # keep_markers subcommand
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
481 keep_markers_parser = subparsers.add_parser('keep_markers', help='Keep markers from JSON file, others will be removed')
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
482 keep_markers_parser.add_argument('--in_json', help="Path to input JSON file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
483 keep_markers_parser.add_argument('--markers', help="Path to file with markers to keep (1 per line)")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
484 keep_markers_parser.add_argument('--out_json', help="Path to output JSON file")
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
485
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
486 args = parser.parse_args()
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
487
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
488 if args.function == 'transform_pkl_to_json':
2
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
489 validate_map_version(Path(args.pkl), 'pkl')
0
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
490 transform_pkl_to_json(Path(args.pkl), Path(args.json))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
491 elif args.function == 'transform_json_to_pkl':
2
44cacbe09bfc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 0966faf8782e9043772acfa32f4a4281687a19dd
iuc
parents: 0
diff changeset
492 validate_map_version(Path(args.json), 'json')
0
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
493 transform_json_to_pkl(Path(args.json), Path(args.pkl))
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
494 elif args.function == 'add_marker':
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
495 add_marker(
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
496 args.in_json,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
497 args.out_json,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
498 args.name,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
499 args.m_length,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
500 args.g_length,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
501 args.gca,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
502 args.k_name,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
503 args.k_id,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
504 args.p_name,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
505 args.p_id,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
506 args.c_name,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
507 args.c_id,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
508 args.o_name,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
509 args.o_id,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
510 args.f_name,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
511 args.f_id,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
512 args.g_name,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
513 args.g_id,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
514 args.s_name,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
515 args.s_id,
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
516 args.t_name)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
517 elif args.function == 'remove_markers':
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
518 remove_markers(args.in_json, args.markers, args.out_json, args.kept_markers)
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
519 elif args.function == 'keep_markers':
506d4a349aa1 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/humann commit 077b8f34e081e6c427acb0fde0fbb97d1b241e0b"
iuc
parents:
diff changeset
520 keep_markers(args.in_json, args.markers, args.out_json)