Mercurial > repos > iss > eurl_vtec_wgs_pt

annotate scripts/ReMatCh/utils/combine_alignment_consensus.py @ 4:32b5625795ea draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 906fafe94520fdbabb0fb94cdbe93ffeab43867a
author iss
date Thu, 19 Oct 2023 19:07:24 +0000
parents c6bab5103a14
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
1 #!/usr/bin/env python3
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
2
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
3 # -*- coding: utf-8 -*-
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
4
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
5 """
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
6 combine_alignment_consensus.py - Combine the alignment consensus
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
7 sequences from ReMatCh first run by reference sequences into single
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
8 files
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
9 <https://github.com/B-UMMI/ReMatCh/>
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
10
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
11 Copyright (C) 2018 Miguel Machado <mpmachado@medicina.ulisboa.pt>
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
12
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
13 Last modified: October 15, 2018
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
14
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
15 This program is free software: you can redistribute it and/or modify
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
16 it under the terms of the GNU General Public License as published by
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
17 the Free Software Foundation, either version 3 of the License, or
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
18 (at your option) any later version.
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
19
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
20 This program is distributed in the hope that it will be useful,
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
21 but WITHOUT ANY WARRANTY; without even the implied warranty of
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
22 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
23 GNU General Public License for more details.
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
24
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
25 You should have received a copy of the GNU General Public License
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
26 along with this program. If not, see <http://www.gnu.org/licenses/>.
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
27 """
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
28
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
29 import os
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
30 import argparse
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
31 import time
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
32 import sys
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
33
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
34 version = '0.2'
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
35
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
36
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
37 def concatenate_files(input_files_list, outdir):
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
38 all_executed_printed = False
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
39 for x, input_file in enumerate(input_files_list):
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
40 sample = os.path.basename(input_file).rsplit('.', 2)[0]
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
41 with open(input_file, 'rtU') as reader:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
42 writer = None
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
43 for line in reader:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
44 line = line.rstrip('\r\n')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
45 if line.startswith('>'):
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
46 file_output = os.path.join(outdir, line[1:] + '.fasta')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
47 if writer is not None:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
48 writer.flush()
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
49 writer.close()
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
50 if os.path.isfile(file_output):
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
51 writer = open(file_output, 'at')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
52 else:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
53 writer = open(file_output, 'wt')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
54 writer.write('>' + sample + '\n')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
55 else:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
56 if len(line) > 0:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
57 writer.write(line + '\n')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
58 writer.flush()
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
59 writer.close()
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
60
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
61 if (x + 1) % 100 == 0:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
62 print('\n' + str(round((float(x + 1) / len(input_files_list)) * 100, 2)) + '% of IDs already processed')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
63 all_executed_printed = True
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
64 if not all_executed_printed:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
65 print('\n' + str(round((float(x + 1) / len(input_files_list)) * 100, 2)) + '% of IDs already processed')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
66
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
67
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
68 def combine_alignment_consensus(args):
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
69 outdir = os.path.abspath(args.outdir)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
70 if not os.path.isdir(outdir):
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
71 os.makedirs(outdir)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
72
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
73 outdir = os.path.join(outdir, 'combine_alignment_consensus_' + time.strftime("%Y%m%d-%H%M%S"), '')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
74 os.makedirs(outdir)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
75
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
76 workdir = os.path.abspath(args.workdir)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
77
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
78 alignment_files = []
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
79 directories = [d for d in os.listdir(workdir) if
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
80 not d.startswith('.') and
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
81 os.path.isdir(os.path.join(workdir, d, ''))]
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
82 for sample_dir in directories:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
83 sample_dir_path = os.path.join(workdir, sample_dir, '')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
84 files = [f for f in os.listdir(sample_dir_path) if
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
85 not f.startswith('.') and
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
86 os.path.isfile(os.path.join(sample_dir_path, f))]
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
87 for file_found in files:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
88 if file_found.endswith('.alignment.fasta'):
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
89 file_found_path = os.path.join(sample_dir_path, file_found)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
90 alignment_files.append(file_found_path)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
91
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
92 if len(alignment_files) > 0:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
93 concatenate_files(alignment_files, outdir)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
94 else:
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
95 sys.exit('No ReMatCh alignment.fasta files were found!')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
96
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
97
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
98 def main():
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
99 parser = argparse.ArgumentParser(prog='combine_alignment_consensus.py',
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
100 description='Combine the alignment consensus sequences from ReMatCh first run by'
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
101 ' reference sequences into single'
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
102 ' files', formatter_class=argparse.ArgumentDefaultsHelpFormatter)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
103 parser.add_argument('--version', help='Version information', action='version', version=str('%(prog)s v' + version))
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
104
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
105 parser_required = parser.add_argument_group('Required options')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
106 parser_required.add_argument('-w', '--workdir', type=str, metavar='/path/to/rematch/working/directory/',
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
107 help='Path to the directory where ReMatCh was running', required=True)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
108
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
109 parser_optional_general = parser.add_argument_group('General facultative options')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
110 parser_optional_general.add_argument('-o', '--outdir', type=str, metavar='/path/to/output/directory/',
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
111 help='Path to the directory where the combined sequence files will stored',
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
112 required=False, default='.')
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
113
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
114 args = parser.parse_args()
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
115
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
116 combine_alignment_consensus(args)
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
117
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
118
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
119 if __name__ == "__main__":
c6bab5103a14 "planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff changeset
120 main()