Mercurial > repos > cstrittmatter > test_eurl_vtec_wgs_pt
annotate scripts/patho_typing.py @ 0:965517909457 draft
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
author | cstrittmatter |
---|---|
date | Wed, 22 Jan 2020 08:41:44 -0500 |
parents | |
children | 0cbed1c0a762 |
rev | line source |
---|---|
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
1 #!/usr/bin/env python |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
2 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
3 # -*- coding: utf-8 -*- |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
4 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
5 """ |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
6 patho_typing.py - In silico pathogenic typing directly from raw |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
7 Illumina reads |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
8 <https://github.com/B-UMMI/patho_typing/> |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
9 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
10 Copyright (C) 2017 Miguel Machado <mpmachado@medicina.ulisboa.pt> |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
11 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
12 Last modified: July 06, 2017 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
13 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
14 This program is free software: you can redistribute it and/or modify |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
15 it under the terms of the GNU General Public License as published by |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
16 the Free Software Foundation, either version 3 of the License, or |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
17 (at your option) any later version. |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
18 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
19 This program is distributed in the hope that it will be useful, |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
20 but WITHOUT ANY WARRANTY; without even the implied warranty of |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
21 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
22 GNU General Public License for more details. |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
23 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
24 You should have received a copy of the GNU General Public License |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
25 along with this program. If not, see <http://www.gnu.org/licenses/>. |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
26 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
27 2017-12-05 ISS |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
28 In order to use the module within the EURL_WGS_Typer pipeline with a |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
29 different virulence database for E coli, mapping against the |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
30 typing_rules.tab was deactivated. |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
31 """ |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
32 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
33 import argparse |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
34 import os |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
35 import time |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
36 import sys |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
37 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
38 import modules.utils as utils |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
39 import modules.run_rematch as run_rematch |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
40 import modules.typing as typing |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
41 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
42 version = '0.3' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
43 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
44 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
45 def set_reference(species, outdir, script_path, trueCoverage): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
46 reference_file = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
47 trueCoverage_file = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
48 trueCoverage_sequences = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
49 trueCoverage_headers = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
50 trueCoverage_config = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
51 typing_file = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
52 typing_sequences = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
53 typing_headers = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
54 typing_rules = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
55 typing_config = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
56 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
57 species_folder = os.path.join(os.path.dirname(script_path), 'modules', 'seq_conf', '_'.join([s.lower() for s in species]), '') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
58 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
59 if os.path.isdir(species_folder): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
60 typing_rules = os.path.join(species_folder, 'typing_rules.tab') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
61 typing_file = os.path.join(species_folder, 'typing.fasta') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
62 typing_sequences, ignore = utils.get_sequence_information(typing_file, 0) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
63 typing_sequences, typing_headers = utils.clean_headers_sequences(typing_sequences) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
64 typing_sequences = utils.simplify_sequence_dict(typing_sequences) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
65 typing_config = os.path.join(species_folder, 'typing.config') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
66 if trueCoverage: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
67 trueCoverage_file = os.path.join(species_folder, 'trueCoverage.fasta') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
68 trueCoverage_sequences, ignore = utils.get_sequence_information(trueCoverage_file, 0) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
69 trueCoverage_sequences, trueCoverage_headers = utils.clean_headers_sequences(trueCoverage_sequences) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
70 trueCoverage_sequences = utils.simplify_sequence_dict(trueCoverage_sequences) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
71 trueCoverage_config = os.path.join(species_folder, 'trueCoverage.config') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
72 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
73 trueCoverage_typing_sequences = trueCoverage_sequences.copy() |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
74 for header in typing_sequences: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
75 if header not in trueCoverage_sequences: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
76 trueCoverage_typing_sequences[header] = typing_sequences[header] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
77 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
78 print 'Sequence {header} of typing.fasta already present in trueCoverage.fasta'.format(header=header) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
79 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
80 reference_file = os.path.join(outdir, 'trueCoverage_typing.fasta') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
81 write_sequeces(reference_file, trueCoverage_typing_sequences) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
82 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
83 reference_file = os.path.join(outdir, 'typing.fasta') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
84 write_sequeces(reference_file, typing_sequences) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
85 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
86 species_present = [] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
87 seq_conf_folder = os.path.join(os.path.dirname(script_path), 'modules', 'seq_conf', '') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
88 species_folder = [d for d in os.listdir(seq_conf_folder) if not d.startswith('.') and os.path.isdir(os.path.join(seq_conf_folder, d, ''))] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
89 for species in species_folder: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
90 species = species.split('_') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
91 species[0] = species[0][0].upper() + species[0][1:] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
92 species_present.append(' '.join(species)) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
93 sys.exit('Only these species are available:' + '\n' + '\n'.join(species_present)) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
94 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
95 return reference_file, trueCoverage_file, trueCoverage_sequences, trueCoverage_headers, trueCoverage_config, typing_file, typing_sequences, typing_headers, typing_rules, typing_config |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
96 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
97 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
98 def confirm_genes_fasta_rules(typing_fasta_headers, typing_rules_file): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
99 with open(typing_rules_file, 'rtU') as reader: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
100 genes = [] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
101 for line in reader: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
102 line = line.splitlines()[0] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
103 if len(line) > 0: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
104 line = line.split('\t') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
105 if line[0].startswith('#'): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
106 genes = line[1:] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
107 break |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
108 missing_genes = [] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
109 for gene in genes: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
110 if gene.lower() not in typing_fasta_headers: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
111 missing_genes.append(gene) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
112 if len(missing_genes) > 0: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
113 sys.exit('The following genes required for typing are not present in typing fasta file: {missing_genes}'.format(missing_genes=', '.join(missing_genes))) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
114 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
115 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
116 def index_fasta_samtools(fasta, region_None, region_outfile_none, print_comand_True): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
117 command = ['samtools', 'faidx', fasta, '', '', ''] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
118 shell_true = False |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
119 if region_None is not None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
120 command[3] = region_None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
121 if region_outfile_none is not None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
122 command[4] = '>' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
123 command[5] = region_outfile_none |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
124 shell_true = True |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
125 run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, shell_true, None, print_comand_True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
126 return run_successfully, stdout |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
127 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
128 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
129 def indexSequenceBowtie2(referenceFile, threads): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
130 if os.path.isfile(str(referenceFile + '.1.bt2')): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
131 run_successfully = True |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
132 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
133 command = ['bowtie2-build', '--threads', str(threads), referenceFile, referenceFile] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
134 run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, False, None, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
135 return run_successfully |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
136 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
137 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
138 def run_bowtie(fastq_files, referenceFile, threads, outdir, conserved_True, numMapLoc): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
139 sam_file = os.path.join(outdir, str('alignment.sam')) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
140 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
141 run_successfully = indexSequenceBowtie2(referenceFile, threads) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
142 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
143 command = ['bowtie2', '-k', str(numMapLoc), '-q', '', '--threads', str(threads), '-x', referenceFile, '', '--no-unal', '-S', sam_file] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
144 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
145 if len(fastq_files) == 1: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
146 command[9] = '-U ' + fastq_files[0] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
147 elif len(fastq_files) == 2: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
148 command[9] = '-1 ' + fastq_files[0] + ' -2 ' + fastq_files[1] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
149 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
150 return False, None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
151 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
152 if conserved_True: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
153 command[4] = '--sensitive' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
154 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
155 command[4] = '--very-sensitive-local' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
156 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
157 run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, False, None, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
158 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
159 if not run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
160 sam_file = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
161 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
162 return run_successfully, sam_file |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
163 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
164 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
165 def sortAlignment(alignment_file, output_file, sortByName_True, threads): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
166 outFormat_string = os.path.splitext(output_file)[1][1:].lower() |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
167 command = ['samtools', 'sort', '-o', output_file, '-O', outFormat_string, '', '-@', str(threads), alignment_file] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
168 if sortByName_True: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
169 command[6] = '-n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
170 run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, False, None, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
171 if not run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
172 output_file = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
173 return run_successfully, output_file |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
174 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
175 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
176 def indexAlignment(alignment_file): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
177 command = ['samtools', 'index', alignment_file] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
178 run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, False, None, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
179 return run_successfully |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
180 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
181 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
182 def mapping_reads(fastq_files, referenceFile, threads, outdir, conserved_True, numMapLoc): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
183 print '\n' + 'Mapping the reads' + '\n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
184 run_successfully, sam_file = run_bowtie(fastq_files, referenceFile, threads, outdir, conserved_True, numMapLoc) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
185 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
186 run_successfully, bam_file = sortAlignment(sam_file, str(os.path.splitext(sam_file)[0] + '.bam'), False, threads) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
187 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
188 os.remove(sam_file) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
189 run_successfully = indexAlignment(bam_file) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
190 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
191 index_fasta_samtools(referenceFile, None, None, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
192 return run_successfully, bam_file |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
193 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
194 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
195 def include_rematch_dependencies_path(): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
196 command = ['which', 'rematch.py'] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
197 run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, False, None, False) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
198 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
199 rematch = stdout.splitlines()[0] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
200 utils.setPATHvariable(False, rematch) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
201 return rematch |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
202 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
203 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
204 def split_bam(bam_file, list_sequences, outdir, threads): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
205 new_bam = os.path.join(outdir, 'partial.bam') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
206 command = ['samtools', 'view', '-b', '-u', '-h', '-o', new_bam, '-@', str(threads), bam_file, ' '.join(list_sequences)] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
207 run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, False, None, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
208 return run_successfully, new_bam |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
209 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
210 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
211 def parse_config(config_file): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
212 config = {'reference_file': None, 'length_extra_seq': None, 'maximum_number_absent_genes': None, 'maximum_number_genes_multiple_alleles': None, 'minimum_read_coverage': None, 'minimum_depth_presence': None, 'minimum_depth_call': None, 'minimum_depth_frequency_dominant_allele': None, 'minimum_gene_coverage': None, 'minimum_gene_identity': None} |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
213 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
214 with open(config_file, 'rtU') as reader: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
215 field = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
216 for line in reader: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
217 line = line.splitlines()[0] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
218 if len(line) > 0: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
219 line = line.split(' ')[0] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
220 if line.startswith('#'): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
221 line = line[1:].split(' ')[0] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
222 field = line |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
223 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
224 if field is not None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
225 if field in ['length_extra_seq', 'maximum_number_absent_genes', 'maximum_number_genes_multiple_alleles', 'minimum_read_coverage', 'minimum_depth_presence', 'minimum_depth_call', 'minimum_gene_coverage', 'minimum_gene_identity']: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
226 line = int(line) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
227 if field in ['minimum_gene_coverage', 'minimum_gene_identity']: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
228 if line < 0 or line > 100: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
229 sys.exit('minimum_gene_coverage in trueCoverage_rematch config file must be an integer between 0 and 100') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
230 elif field == 'minimum_depth_frequency_dominant_allele': |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
231 line = float(line) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
232 if line < 0 or line > 1: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
233 sys.exit('minimum_depth_frequency_dominant_allele in trueCoverage_rematch config file must be a double between 0 and 1') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
234 config[field] = line |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
235 field = None |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
236 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
237 for field in config: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
238 if config[field] is None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
239 sys.exit(field + ' in trueCoverage_rematch config file is missing') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
240 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
241 return config |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
242 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
243 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
244 def clean_pathotyping_folder(outdir, reference_file, debug_mode_true): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
245 if not debug_mode_true: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
246 files = [f for f in os.listdir(outdir) if not f.startswith('.') and os.path.isfile(os.path.join(outdir, f))] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
247 for file_found in files: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
248 if file_found.startswith(('alignment.', os.path.splitext(os.path.basename(reference_file))[0])): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
249 file_found = os.path.join(outdir, file_found) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
250 os.remove(file_found) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
251 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
252 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
253 def write_sequeces(out_file, sequences_dict): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
254 with open(out_file, 'wt') as writer: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
255 for header in sequences_dict: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
256 writer.write('>' + header + '\n') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
257 writer.write('\n'.join(utils.chunkstring(sequences_dict[header]['sequence'], 80)) + '\n') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
258 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
259 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
260 def main(): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
261 parser = argparse.ArgumentParser(prog='patho_typing.py', description='In silico pathogenic typing directly from raw Illumina reads', formatter_class=argparse.ArgumentDefaultsHelpFormatter) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
262 parser.add_argument('--version', help='Version information', action='version', version=str('%(prog)s v' + version)) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
263 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
264 parser_required = parser.add_argument_group('Required options') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
265 parser_required.add_argument('-f', '--fastq', nargs='+', action=utils.required_length((1, 2), '--fastq'), type=argparse.FileType('r'), metavar=('/path/to/input/file.fq.gz'), help='Path to single OR paired-end fastq files. If two files are passed, they will be assumed as being the paired fastq files', required=True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
266 parser_required.add_argument('-s', '--species', nargs=2, type=str, metavar=('Yersinia', 'enterocolitica'), help='Species name', required=True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
267 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
268 parser_optional_general = parser.add_argument_group('General facultative options') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
269 parser_optional_general.add_argument('-o', '--outdir', type=str, metavar='/path/to/output/directory/', help='Path to the directory where the information will be stored', required=False, default='.') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
270 parser_optional_general.add_argument('-j', '--threads', type=int, metavar='N', help='Number of threads to use', required=False, default=1) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
271 parser_optional_general.add_argument('--trueCoverage', action='store_true', help='Assess true coverage before continue typing') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
272 parser_optional_general.add_argument('--noCheckPoint', action='store_true', help='Ignore the true coverage checking point') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
273 parser_optional_general.add_argument('--minGeneCoverage', type=int, metavar='N', help='Minimum typing percentage of target reference gene sequence covered to consider a gene to be present (value between [0, 100])', required=False) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
274 parser_optional_general.add_argument('--minGeneIdentity', type=int, metavar='N', help='Minimum typing percentage of identity of reference gene sequence covered to consider a gene to be present (value between [0, 100]). One INDEL will be considered as one difference', required=False) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
275 parser_optional_general.add_argument('--minGeneDepth', type=int, metavar='N', help='Minimum typing gene average coverage depth of present positions to consider a gene to be present (default 15, or 1/3 of average sample coverage assessed by true coverage analysis)', required=False) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
276 parser_optional_general.add_argument('--doNotRemoveConsensus', action='store_true', help='Do not remove ReMatCh consensus sequences') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
277 parser_optional_general.add_argument('--debug', action='store_true', help='DeBug Mode: do not remove temporary files') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
278 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
279 args = parser.parse_args() |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
280 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
281 if args.minGeneCoverage is not None and (args.minGeneCoverage < 0 or args.minGeneCoverage > 100): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
282 parser.error('--minGeneCoverage should be a value between [0, 100]') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
283 if args.minGeneIdentity is not None and (args.minGeneIdentity < 0 or args.minGeneIdentity > 100): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
284 parser.error('--minGeneIdentity should be a value between [0, 100]') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
285 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
286 start_time = time.time() |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
287 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
288 args.outdir = os.path.abspath(args.outdir) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
289 if not os.path.isdir(args.outdir): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
290 os.makedirs(args.outdir) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
291 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
292 # Start logger |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
293 logfile, time_str = utils.start_logger(args.outdir) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
294 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
295 script_path = utils.general_information(logfile, version, args.outdir, time_str) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
296 print '\n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
297 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
298 rematch = include_rematch_dependencies_path() |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
299 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
300 args.fastq = [fastq.name for fastq in args.fastq] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
301 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
302 reference_file, trueCoverage_file, trueCoverage_sequences, trueCoverage_headers, trueCoverage_config, typing_file, typing_sequences, typing_headers, typing_rules, typing_config = set_reference(args.species, args.outdir, script_path, args.trueCoverage) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
303 original_reference_file = str(reference_file) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
304 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
305 #confirm_genes_fasta_rules(typing_headers, typing_rules) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
306 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
307 run_successfully, bam_file = mapping_reads(args.fastq, reference_file, args.threads, args.outdir, False, 1) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
308 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
309 rematch_dir = os.path.join(args.outdir, 'rematch', '') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
310 if not os.path.isdir(rematch_dir): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
311 os.makedirs(rematch_dir) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
312 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
313 if args.trueCoverage: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
314 trueCoverage_dir = os.path.join(rematch_dir, 'trueCoverage', '') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
315 if not os.path.isdir(trueCoverage_dir): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
316 os.makedirs(trueCoverage_dir) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
317 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
318 print '\n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
319 run_successfully, trueCoverage_bam = split_bam(bam_file, trueCoverage_headers, trueCoverage_dir, args.threads) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
320 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
321 run_successfully = indexAlignment(trueCoverage_bam) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
322 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
323 reference_file = os.path.join(trueCoverage_dir, 'reference.fasta') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
324 write_sequeces(reference_file, trueCoverage_sequences) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
325 index_fasta_samtools(reference_file, None, None, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
326 config = parse_config(trueCoverage_config) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
327 runtime, run_successfully, sample_data_general, data_by_gene = run_rematch.run_rematch(rematch, trueCoverage_dir, reference_file, trueCoverage_bam, args.threads, config['length_extra_seq'], config['minimum_depth_presence'], config['minimum_depth_call'], config['minimum_depth_frequency_dominant_allele'], config['minimum_gene_coverage'], config['minimum_gene_identity'], args.debug, args.doNotRemoveConsensus) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
328 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
329 if run_successfully and sample_data_general['mean_sample_coverage'] is not None and sample_data_general['number_absent_genes'] is not None and sample_data_general['number_genes_multiple_alleles'] is not None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
330 if args.minGeneDepth is None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
331 args.minGeneDepth = sample_data_general['mean_sample_coverage'] / 3 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
332 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
333 exit_info = [] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
334 if sample_data_general['mean_sample_coverage'] < config['minimum_read_coverage']: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
335 exit_info.append('Sample coverage ({mean_sample_coverage}) lower than the minimum required ({minimum_read_coverage})'.format(mean_sample_coverage=sample_data_general['mean_sample_coverage'], minimum_read_coverage=config['minimum_read_coverage'])) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
336 if sample_data_general['number_absent_genes'] > config['maximum_number_absent_genes']: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
337 exit_info.append('Number of absent genes ({number_absent_genes}) higher than the maximum allowed ({maximum_number_absent_genes})'.format(number_absent_genes=sample_data_general['number_absent_genes'], maximum_number_absent_genes=config['maximum_number_absent_genes'])) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
338 if sample_data_general['number_genes_multiple_alleles'] > config['maximum_number_genes_multiple_alleles']: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
339 exit_info.append('Number of genes with multiple alleles ({number_genes_multiple_alleles}) higher than the maximum allowed ({maximum_number_genes_multiple_alleles})'.format(number_genes_multiple_alleles=sample_data_general['number_genes_multiple_alleles'], maximum_number_genes_multiple_alleles=config['maximum_number_genes_multiple_alleles'])) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
340 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
341 if len(exit_info) > 0: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
342 print '\n' + '\n'.join(exit_info) + '\n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
343 e = 'TrueCoverage requirements not fulfilled' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
344 print '\n' + e + '\n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
345 if not args.noCheckPoint: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
346 clean_pathotyping_folder(args.outdir, original_reference_file, args.debug) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
347 time_taken = utils.runTime(start_time) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
348 sys.exit(e) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
349 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
350 e = 'TrueCoverage module did not run successfully' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
351 print '\n' + e + '\n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
352 if not args.noCheckPoint: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
353 clean_pathotyping_folder(args.outdir, original_reference_file, args.debug) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
354 time_taken = utils.runTime(start_time) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
355 sys.exit(e) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
356 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
357 print '\n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
358 typing_dir = os.path.join(rematch_dir, 'typing', '') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
359 if not os.path.isdir(typing_dir): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
360 os.makedirs(typing_dir) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
361 run_successfully, bam_file = split_bam(bam_file, typing_headers, typing_dir, args.threads) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
362 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
363 run_successfully = indexAlignment(bam_file) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
364 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
365 reference_file = os.path.join(typing_dir, 'reference.fasta') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
366 write_sequeces(reference_file, typing_sequences) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
367 index_fasta_samtools(reference_file, None, None, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
368 rematch_dir = str(typing_dir) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
369 if not run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
370 if args.noCheckPoint: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
371 clean_pathotyping_folder(args.outdir, original_reference_file, args.debug) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
372 time_taken = utils.runTime(start_time) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
373 sys.exit('Something in the required TrueCoverage analysis went wrong') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
374 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
375 if run_successfully: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
376 config = parse_config(typing_config) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
377 if args.minGeneCoverage is not None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
378 config['minimum_gene_coverage'] = args.minGeneCoverage |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
379 if args.minGeneIdentity is not None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
380 config['minimum_gene_identity'] = args.minGeneIdentity |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
381 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
382 runtime, run_successfully, sample_data_general, data_by_gene = run_rematch.run_rematch(rematch, rematch_dir, reference_file, bam_file, args.threads, config['length_extra_seq'], config['minimum_depth_presence'], config['minimum_depth_call'], config['minimum_depth_frequency_dominant_allele'], config['minimum_gene_coverage'], config['minimum_gene_identity'], args.debug, args.doNotRemoveConsensus) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
383 if run_successfully and data_by_gene is not None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
384 if args.minGeneDepth is None: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
385 args.minGeneDepth = 15 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
386 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
387 #runtime, ignore, ignore = typing.typing(data_by_gene, typing_rules, config['minimum_gene_coverage'], config['minimum_gene_identity'], args.minGeneDepth, args.outdir) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
388 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
389 clean_pathotyping_folder(args.outdir, original_reference_file, args.debug) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
390 time_taken = utils.runTime(start_time) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
391 sys.exit('ReMatCh run for pathotyping did not run successfully') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
392 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
393 clean_pathotyping_folder(args.outdir, original_reference_file, args.debug) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
394 time_taken = utils.runTime(start_time) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
395 sys.exit('Something did not run successfully') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
396 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
397 clean_pathotyping_folder(args.outdir, original_reference_file, args.debug) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
398 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
399 print '\n' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
400 time_taken = utils.runTime(start_time) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
401 del time_taken |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
402 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
403 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
404 if __name__ == "__main__": |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
405 main() |