annotate data_manager/hisat_index_builder.py @ 0:8082f04de7ae draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
author devteam
date Fri, 06 Nov 2015 14:17:24 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
1 #!/usr/bin/env python
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
2 # Based heavily on the Bowtie 2 data manager wrapper script by Dan Blankenberg
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
3
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
4 import sys
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
5 import os
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
6 import optparse
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
7 import subprocess
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
8
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
9 from json import loads, dumps
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
10
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
11
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
12 DEFAULT_DATA_TABLE_NAME = "hisat_indexes"
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
13
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
14 def get_id_name( params, dbkey, fasta_description=None):
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
15 #TODO: ensure sequence_id is unique and does not already appear in location file
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
16 sequence_id = params['param_dict']['sequence_id']
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
17 if not sequence_id:
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
18 sequence_id = dbkey
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
19
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
20 sequence_name = params['param_dict']['sequence_name']
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
21 if not sequence_name:
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
22 sequence_name = fasta_description
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
23 if not sequence_name:
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
24 sequence_name = dbkey
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
25 return sequence_id, sequence_name
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
26
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
27 def build_hisat_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME ):
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
28 #TODO: allow multiple FASTA input files
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
29 fasta_base_name = os.path.split( fasta_filename )[-1]
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
30 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
31 os.symlink( fasta_filename, sym_linked_fasta_filename )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
32 args = [ 'hisat-build', sym_linked_fasta_filename, sequence_id ]
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
33 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
34 return_code = proc.wait()
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
35 if return_code:
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
36 print >> sys.stderr, "Error building index."
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
37 sys.exit( return_code )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
38 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=sequence_id )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
39 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
40
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
41 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
42 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
43 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
44 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
45 return data_manager_dict
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
46
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
47 def main():
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
48 #Parse Command Line
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
49 parser = optparse.OptionParser()
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
50 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
51 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
52 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
53 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
54 (options, args) = parser.parse_args()
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
55
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
56 filename = args[0]
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
57
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
58 params = loads( open( filename ).read() )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
59 target_directory = params[ 'output_data' ][0]['extra_files_path']
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
60 os.mkdir( target_directory )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
61 data_manager_dict = {}
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
62
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
63 dbkey = options.fasta_dbkey
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
64
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
65 if dbkey in [ None, '', '?' ]:
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
66 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
67
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
68 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
69
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
70 #build the index
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
71 build_hisat_index( data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
72
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
73 #save info to json file
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
74 open( filename, 'wb' ).write( dumps( data_manager_dict ) )
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
75
8082f04de7ae planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_hisat_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
76 if __name__ == "__main__": main()