Mercurial > repos > devteam > data_manager_sam_fasta_index_builder
annotate data_manager/data_manager_sam_fasta_index_builder.py @ 8:3ec5c51384e6 draft
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 20381ede5940c6bbca00b550e01550f5aad0f2b7"
author | iuc |
---|---|
date | Mon, 01 Nov 2021 08:32:15 +0000 |
parents | 3e4d45aca422 |
children |
rev | line source |
---|---|
0 | 1 #!/usr/bin/env python |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
2 # Dan Blankenberg |
0 | 3 |
4 import json | |
5 import optparse | |
6 import os | |
7 import subprocess | |
8 import sys | |
9 import tempfile | |
10 | |
11 CHUNK_SIZE = 2**20 | |
12 | |
13 DEFAULT_DATA_TABLE_NAME = "fasta_indexes" | |
14 | |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
15 |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
16 def get_id_name(params, dbkey, fasta_description=None): |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
17 # TODO: ensure sequence_id is unique and does not already appear in location file |
0 | 18 sequence_id = params['param_dict']['sequence_id'] |
19 if not sequence_id: | |
20 sequence_id = dbkey | |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
21 |
0 | 22 sequence_name = params['param_dict']['sequence_name'] |
23 if not sequence_name: | |
24 sequence_name = fasta_description | |
25 if not sequence_name: | |
26 sequence_name = dbkey | |
27 return sequence_id, sequence_name | |
28 | |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
29 |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
30 def build_sam_index(data_manager_dict, fasta_filename, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME): |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
31 # TODO: allow multiple FASTA input files |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
32 assert os.path.exists(fasta_filename), 'FASTA file "%s" is missing, cannot build samtools index.' % fasta_filename |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
33 fasta_base_name = os.path.split(fasta_filename)[-1] |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
34 sym_linked_fasta_filename = os.path.join(target_directory, fasta_base_name) |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
35 os.symlink(fasta_filename, sym_linked_fasta_filename) |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
36 |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
37 args = ['samtools', 'faidx'] |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
38 args.append(sym_linked_fasta_filename) |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
39 tmp_stderr = tempfile.NamedTemporaryFile(prefix="tmp-data-manager-sam_fa_index_builder-stderr") |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
40 proc = subprocess.Popen(args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno()) |
0 | 41 return_code = proc.wait() |
42 if return_code: | |
43 tmp_stderr.flush() | |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
44 tmp_stderr.seek(0) |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
45 sys.stderr.write("Error building index:\n") |
0 | 46 while True: |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
47 chunk = tmp_stderr.read(CHUNK_SIZE) |
0 | 48 if not chunk: |
49 break | |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
50 sys.stderr.write(chunk) |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
51 sys.exit(return_code) |
0 | 52 tmp_stderr.close() |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
53 data_table_entry = dict(value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name) |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
54 _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry) |
0 | 55 |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
56 |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
57 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry): |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
58 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
59 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, []) |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
60 data_manager_dict['data_tables'][data_table_name].append(data_table_entry) |
0 | 61 return data_manager_dict |
62 | |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
63 |
0 | 64 def main(): |
65 parser = optparse.OptionParser() | |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
66 parser.add_option('-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename') |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
67 parser.add_option('-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey') |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
68 parser.add_option('-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description') |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
69 parser.add_option('-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name') |
0 | 70 (options, args) = parser.parse_args() |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
71 |
0 | 72 filename = args[0] |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
73 |
7
3e4d45aca422
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
5
diff
changeset
|
74 with open(filename) as fh: |
3e4d45aca422
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
5
diff
changeset
|
75 params = json.load(fh) |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
76 target_directory = params['output_data'][0]['extra_files_path'] |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
77 os.mkdir(target_directory) |
0 | 78 data_manager_dict = {} |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
79 |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
80 if options.fasta_dbkey in [None, '', '?']: |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
81 raise Exception('"%s" is not a valid dbkey. You must specify a valid dbkey.' % (options.fasta_dbkey)) |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
82 |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
83 sequence_id, sequence_name = get_id_name(params, dbkey=options.fasta_dbkey, fasta_description=options.fasta_description) |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
84 |
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
85 # build the index |
5
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
86 build_sam_index( |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
87 data_manager_dict, |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
88 options.fasta_filename, |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
89 target_directory, |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
90 options.fasta_dbkey, |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
91 sequence_id, |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
92 sequence_name, |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
93 data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME |
cc4ef4d38cf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 72fb3d356a87d2b555572486e7901449c3bdb568"
iuc
parents:
3
diff
changeset
|
94 ) |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
95 |
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
96 # save info to json file |
7
3e4d45aca422
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
5
diff
changeset
|
97 with open(filename, 'w') as fh: |
3e4d45aca422
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
5
diff
changeset
|
98 json.dump(data_manager_dict, fh, sort_keys=True) |
3
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
99 |
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
100 |
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
101 if __name__ == "__main__": |
1f54e98616af
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
102 main() |