comparison data_manager/picard_index_builder.py @ 0:9a024421602a draft

Uploaded
author devteam
date Thu, 19 Mar 2015 16:41:17 -0400
parents
children 0663d25e619c
comparison
equal deleted inserted replaced
-1:000000000000 0:9a024421602a
1 #!/usr/bin/env python
2 #Dan Blankenberg
3
4 import sys
5 import os
6 import optparse
7 import subprocess
8
9 from json import loads, dumps
10
11
12 DEFAULT_DATA_TABLE_NAME = "picard_indexes"
13
14 def get_id_name( params, dbkey, fasta_description=None):
15 #TODO: ensure sequence_id is unique and does not already appear in location file
16 sequence_id = params['param_dict']['sequence_id']
17 if not sequence_id:
18 sequence_id = dbkey
19
20 sequence_name = params['param_dict']['sequence_name']
21 if not sequence_name:
22 sequence_name = fasta_description
23 if not sequence_name:
24 sequence_name = dbkey
25 return sequence_id, sequence_name
26
27 def build_picard_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME ):
28 #TODO: allow multiple FASTA input files
29 fasta_base_name = os.path.split( fasta_filename )[-1]
30 prefix = os.path.splitext( fasta_base_name )[0]
31 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
32 os.symlink( fasta_filename, sym_linked_fasta_filename )
33 args = [ 'java', '-Xmx4G', '-jar', os.path.join( os.environ[ 'JAVA_JAR_PATH' ], 'CreateSequenceDictionary.jar' ), 'R=%s' % sym_linked_fasta_filename, 'O=%s.dict' % prefix ]
34 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory )
35 return_code = proc.wait()
36 if return_code:
37 print >> sys.stderr, "Error building index."
38 sys.exit( return_code )
39 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
40 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
41
42 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
43 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
44 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
45 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
46 return data_manager_dict
47
48 def main():
49 #Parse Command Line
50 parser = optparse.OptionParser()
51 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
52 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
53 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
54 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
55 (options, args) = parser.parse_args()
56
57 filename = args[0]
58
59 params = loads( open( filename ).read() )
60 target_directory = params[ 'output_data' ][0]['extra_files_path']
61 os.mkdir( target_directory )
62 data_manager_dict = {}
63
64 dbkey = options.fasta_dbkey
65
66 if dbkey in [ None, '', '?' ]:
67 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
68
69 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
70
71 #build the index
72 build_picard_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 )
73
74 #save info to json file
75 open( filename, 'wb' ).write( dumps( data_manager_dict ) )
76
77 if __name__ == "__main__": main()