comparison data_manager/bwa_mem_index_builder.py @ 0:36447bb36384 draft

Uploaded
author devteam
date Wed, 18 Mar 2015 15:18:49 -0400
parents
children 46066df8813d
comparison
equal deleted inserted replaced
-1:000000000000 0:36447bb36384
1 #!/usr/bin/env python
2 #Dan Blankenberg
3
4 import sys
5 import os
6 import tempfile
7 import optparse
8 import subprocess
9
10 from json import loads, dumps
11
12
13 CHUNK_SIZE = 2**20
14 TWO_GB = 2**30*2
15
16 DEFAULT_DATA_TABLE_NAME = "bwa_mem_indexes"
17
18 def get_id_name( params, dbkey, fasta_description=None):
19 #TODO: ensure sequence_id is unique and does not already appear in location file
20 sequence_id = params['param_dict']['sequence_id']
21 if not sequence_id:
22 sequence_id = dbkey
23
24 sequence_name = params['param_dict']['sequence_name']
25 if not sequence_name:
26 sequence_name = fasta_description
27 if not sequence_name:
28 sequence_name = dbkey
29 return sequence_id, sequence_name
30
31 def build_bwa_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME ):
32 #TODO: allow multiple FASTA input files
33 #tmp_dir = tempfile.mkdtemp( prefix='tmp-data-manager-bwa-index-builder-' )
34 fasta_base_name = os.path.split( fasta_filename )[-1]
35 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
36 os.symlink( fasta_filename, sym_linked_fasta_filename )
37 if params['param_dict']['index_algorithm'] == 'automatic':
38 if os.stat( fasta_filename ).st_size < TWO_GB: #use 2 GB as cut off for memory vs. max of 2gb database size; this is somewhat arbitrary
39 index_algorithm = 'is'
40 else:
41 index_algorithm = 'bwtsw'
42 else:
43 index_algorithm = params['param_dict']['index_algorithm']
44
45 args = [ 'bwa', 'index', '-a', index_algorithm ]
46 args.append( sym_linked_fasta_filename )
47 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory )
48 return_code = proc.wait()
49 if return_code:
50 print >> sys.stderr, "Error building index."
51 sys.exit( return_code )
52 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
53 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
54
55 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
56 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
57 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
58 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
59 return data_manager_dict
60
61 def main():
62 #Parse Command Line
63 parser = optparse.OptionParser()
64 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
65 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
66 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
67 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
68 (options, args) = parser.parse_args()
69
70 filename = args[0]
71
72 params = loads( open( filename ).read() )
73 target_directory = params[ 'output_data' ][0]['extra_files_path']
74 os.mkdir( target_directory )
75 data_manager_dict = {}
76
77 dbkey = options.fasta_dbkey
78
79 if dbkey in [ None, '', '?' ]:
80 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
81
82 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
83
84 #build the index
85 build_bwa_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 )
86
87 #save info to json file
88 open( filename, 'wb' ).write( dumps( data_manager_dict ) )
89
90 if __name__ == "__main__": main()