annotate data_manager/data_manager_sam_fasta_index_builder.py @ 3:1f54e98616af draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
author iuc
date Tue, 04 Apr 2017 18:13:56 -0400
parents 2a1ac1abc3f7
children cc4ef4d38cf9
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
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
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
3
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
4 import json
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
5 import optparse
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
6 import os
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
7 import subprocess
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
8 import sys
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
9 import tempfile
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
10
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
11 CHUNK_SIZE = 2**20
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
12
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
13 DEFAULT_DATA_TABLE_NAME = "fasta_indexes"
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
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
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
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
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
18 sequence_id = params['param_dict']['sequence_id']
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
19 if not sequence_id:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
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
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
22 sequence_name = params['param_dict']['sequence_name']
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
23 if not sequence_name:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
24 sequence_name = fasta_description
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
25 if not sequence_name:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
26 sequence_name = dbkey
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
27 return sequence_id, sequence_name
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
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
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
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
2
2a1ac1abc3f7 planemo upload commit 5ad726dc73203a704666033cd3bf70b82575978f
devteam
parents: 0
diff changeset
32 assert os.path.exists( fasta_filename ), 'FASTA file "%s" is missing, cannot build samtools index.' % fasta_filename
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
33 fasta_base_name = os.path.split( fasta_filename )[-1]
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
34 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
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
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
37 args = [ 'samtools', 'faidx' ]
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
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")
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
40 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
41 return_code = proc.wait()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
42 if return_code:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
43 tmp_stderr.flush()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
44 tmp_stderr.seek( 0 )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
45 sys.stderr.write( "Error building index:\n" )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
46 while True:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
47 chunk = tmp_stderr.read( CHUNK_SIZE )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
48 if not chunk:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
49 break
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
50 sys.stderr.write( chunk )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
51 sys.exit( return_code )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
52 tmp_stderr.close()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
53 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
54 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
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
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
57 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
58 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
59 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
60 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
61 return data_manager_dict
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
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
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
64 def main():
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
65 parser = optparse.OptionParser()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
66 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
67 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
68 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
69 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
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
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
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
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
74 params = json.loads( open( filename ).read() )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
75 target_directory = params[ 'output_data' ][0]['extra_files_path']
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
76 os.mkdir( target_directory )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
77 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
78
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
79 if options.fasta_dbkey in [ None, '', '?' ]:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
80 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
81
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
82 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
83
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 # build the index
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
85 build_sam_index( data_manager_dict, options.fasta_filename, target_directory, options.fasta_dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name or 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
86
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
87 # save info to json file
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
88 open( filename, 'wb' ).write( json.dumps( 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
89
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
90
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
91 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
92 main()