annotate data_manager/picard_index_builder.py @ 6:fa8c605bf385 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 096286097ed5cdf189a1b68c3fc34d10f4142e54
author iuc
date Sun, 16 Apr 2023 08:28:16 +0000
parents acaa6719a478
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9a024421602a Uploaded
devteam
parents:
diff changeset
1 #!/usr/bin/env python
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
2 # Dan Blankenberg
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
3 from __future__ import print_function
0
9a024421602a Uploaded
devteam
parents:
diff changeset
4
6
fa8c605bf385 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 096286097ed5cdf189a1b68c3fc34d10f4142e54
iuc
parents: 5
diff changeset
5 import json
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
6 import optparse
0
9a024421602a Uploaded
devteam
parents:
diff changeset
7 import os
9a024421602a Uploaded
devteam
parents:
diff changeset
8 import subprocess
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
9 import sys
0
9a024421602a Uploaded
devteam
parents:
diff changeset
10
9a024421602a Uploaded
devteam
parents:
diff changeset
11 DEFAULT_DATA_TABLE_NAME = "picard_indexes"
9a024421602a Uploaded
devteam
parents:
diff changeset
12
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
13
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
14 def get_id_name(params, dbkey, fasta_description=None):
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
15 # TODO: ensure sequence_id is unique and does not already appear in location file
0
9a024421602a Uploaded
devteam
parents:
diff changeset
16 sequence_id = params['param_dict']['sequence_id']
9a024421602a Uploaded
devteam
parents:
diff changeset
17 if not sequence_id:
9a024421602a Uploaded
devteam
parents:
diff changeset
18 sequence_id = dbkey
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
19
0
9a024421602a Uploaded
devteam
parents:
diff changeset
20 sequence_name = params['param_dict']['sequence_name']
9a024421602a Uploaded
devteam
parents:
diff changeset
21 if not sequence_name:
9a024421602a Uploaded
devteam
parents:
diff changeset
22 sequence_name = fasta_description
9a024421602a Uploaded
devteam
parents:
diff changeset
23 if not sequence_name:
9a024421602a Uploaded
devteam
parents:
diff changeset
24 sequence_name = dbkey
9a024421602a Uploaded
devteam
parents:
diff changeset
25 return sequence_id, sequence_name
9a024421602a Uploaded
devteam
parents:
diff changeset
26
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
27
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
28 def build_picard_index(data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME):
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
29 # TODO: allow multiple FASTA input files
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
30 fasta_base_name = os.path.split(fasta_filename)[-1]
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
31 prefix = os.path.splitext(fasta_base_name)[0]
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
32 sym_linked_fasta_filename = os.path.join(target_directory, fasta_base_name)
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
33 os.symlink(fasta_filename, sym_linked_fasta_filename)
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
34 args = ['picard', 'CreateSequenceDictionary', 'R=%s' % sym_linked_fasta_filename, 'O=%s.dict' % prefix]
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
35 proc = subprocess.Popen(args=args, shell=False, cwd=target_directory)
0
9a024421602a Uploaded
devteam
parents:
diff changeset
36 return_code = proc.wait()
9a024421602a Uploaded
devteam
parents:
diff changeset
37 if return_code:
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
38 print("Error building index.", file=sys.stderr)
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
39 sys.exit(return_code)
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
40 data_table_entry = dict(value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name)
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
41 _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry)
0
9a024421602a Uploaded
devteam
parents:
diff changeset
42
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
43
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
44 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry):
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
45 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {})
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
46 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, [])
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
47 data_manager_dict['data_tables'][data_table_name].append(data_table_entry)
0
9a024421602a Uploaded
devteam
parents:
diff changeset
48 return data_manager_dict
9a024421602a Uploaded
devteam
parents:
diff changeset
49
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
50
0
9a024421602a Uploaded
devteam
parents:
diff changeset
51 def main():
9a024421602a Uploaded
devteam
parents:
diff changeset
52 parser = optparse.OptionParser()
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
53 parser.add_option('-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename')
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
54 parser.add_option('-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey')
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
55 parser.add_option('-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description')
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
56 parser.add_option('-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name')
0
9a024421602a Uploaded
devteam
parents:
diff changeset
57 (options, args) = parser.parse_args()
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
58
0
9a024421602a Uploaded
devteam
parents:
diff changeset
59 filename = args[0]
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
60
6
fa8c605bf385 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 096286097ed5cdf189a1b68c3fc34d10f4142e54
iuc
parents: 5
diff changeset
61 with open(filename) as fh:
fa8c605bf385 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 096286097ed5cdf189a1b68c3fc34d10f4142e54
iuc
parents: 5
diff changeset
62 params = json.load(fh)
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
63 target_directory = params['output_data'][0]['extra_files_path']
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
64 os.mkdir(target_directory)
0
9a024421602a Uploaded
devteam
parents:
diff changeset
65 data_manager_dict = {}
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
66
0
9a024421602a Uploaded
devteam
parents:
diff changeset
67 dbkey = options.fasta_dbkey
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
68
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
69 if dbkey in [None, '', '?']:
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
70 raise Exception('"%s" is not a valid dbkey. You must specify a valid dbkey.' % (dbkey))
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
71
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
72 sequence_id, sequence_name = get_id_name(params, dbkey=dbkey, fasta_description=options.fasta_description)
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
73
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
74 # build the index
5
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
75 build_picard_index(data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id,
acaa6719a478 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 31fc4022a8c2fef50a3aff9210091be21e8450ef"
iuc
parents: 4
diff changeset
76 sequence_name, data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME)
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
77
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
78 # save info to json file
6
fa8c605bf385 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 096286097ed5cdf189a1b68c3fc34d10f4142e54
iuc
parents: 5
diff changeset
79 with open(filename, 'w') as fh:
fa8c605bf385 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 096286097ed5cdf189a1b68c3fc34d10f4142e54
iuc
parents: 5
diff changeset
80 json.dump(data_manager_dict, fh, sort_keys=True)
3
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
81
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
82
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
83 if __name__ == "__main__":
0663d25e619c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
84 main()