comparison data_manager/bowtie2_index_builder.py @ 0:9458af1ab90b draft

Uploaded
author devteam
date Tue, 14 Apr 2015 12:06:01 -0400
parents
children 9bf4eb559ed5
comparison
equal deleted inserted replaced
-1:000000000000 0:9458af1ab90b
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 = "bowtie2_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_bowtie2_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 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
31 os.symlink( fasta_filename, sym_linked_fasta_filename )
32 args = [ 'bowtie2-build', sym_linked_fasta_filename, sequence_id ]
33 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory )
34 return_code = proc.wait()
35 if return_code:
36 print >> sys.stderr, "Error building index."
37 sys.exit( return_code )
38 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=sequence_id )
39 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
40
41 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
42 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
43 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
44 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
45 return data_manager_dict
46
47 def main():
48 #Parse Command Line
49 parser = optparse.OptionParser()
50 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
51 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
52 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
53 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
54 (options, args) = parser.parse_args()
55
56 filename = args[0]
57
58 params = loads( open( filename ).read() )
59 target_directory = params[ 'output_data' ][0]['extra_files_path']
60 os.mkdir( target_directory )
61 data_manager_dict = {}
62
63 dbkey = options.fasta_dbkey
64
65 if dbkey in [ None, '', '?' ]:
66 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
67
68 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
69
70 #build the index
71 build_bowtie2_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 )
72
73 #save info to json file
74 open( filename, 'wb' ).write( dumps( data_manager_dict ) )
75
76 if __name__ == "__main__": main()