comparison data_manager/rnastar_index_builder.py @ 0:84f6e67cbae5 draft

Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
author fubar
date Mon, 29 Sep 2014 20:25:10 -0400
parents
children ebadd2c92958
comparison
equal deleted inserted replaced
-1:000000000000 0:84f6e67cbae5
1 #!/usr/bin/env python
2 #Dan Blankenberg
3 # adapted from Dan's BWA one for rna star
4 # ross lazarus sept 2014
5 #
6 import sys
7 import os
8 import tempfile
9 import optparse
10 import subprocess
11
12 from galaxy.util.json import from_json_string, to_json_string
13
14 DEFAULT_DATA_TABLE_NAME = "rnastar_indexes"
15
16 def get_id_name( params, dbkey, fasta_description=None):
17 #TODO: ensure sequence_id is unique and does not already appear in location file
18 sequence_id = params['param_dict']['sequence_id']
19 if not sequence_id:
20 sequence_id = dbkey
21
22 sequence_name = params['param_dict']['sequence_name']
23 if not sequence_name:
24 sequence_name = fasta_description
25 if not sequence_name:
26 sequence_name = dbkey
27 return sequence_id, sequence_name
28
29 def build_rnastar_index( data_manager_dict, fasta_filename, target_directory, dbkey, sequence_id, sequence_name, data_table_name,
30 sjdbOverhang,sjdbGTFfile, sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript,sjdbGTFfeatureExon,sjdbGTFchrPrefix,n_threads):
31 #TODO: allow multiple FASTA input files
32 #tmp_dir = tempfile.mkdtemp( prefix='tmp-data-manager-bwa-index-builder-' )
33 fasta_base_name = os.path.split( fasta_filename )[-1]
34 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
35 os.symlink( fasta_filename, sym_linked_fasta_filename )
36 pdict={'target_directory':target_directory,'n_threads':n_threads, 'sjdbFileChrStartEnd':sjdbFileChrStartEnd,
37 'sjdbGTFtagExonParentTranscript':sjdbGTFtagExonParentTranscript, 'sjdbGTFfeatureExon':sjdbGTFfeatureExon,
38 'sjdbGTFchrPrefix':sjdbGTFchrPrefix,'sjdbOverhang':sjdbOverhang, 'sjdbGTFfile':sjdbGTFfile,
39 'sym_linked_fasta_filename':sym_linked_fasta_filename}
40
41 cl = 'STAR --runMode genomeGenerate --genomeFastaFiles %(sym_linked_fasta_filename)s --genomeDir %(target_directory)s --runThreadN %(n_threads)s' % pdict
42 if sjdbGTFfile:
43 cl += '''--sjdbGTFchrPrefix %(sjdbGTFchrPrefix)s --sjdbGTFfeatureExon %(sjdbGTFfeatureExon)s --sjdbOverhang %(sjdbOverhang)s
44 --sjdbGTFfile %(sjdbGTFfile)s --sjdbGTFtagExonParentTranscript %(sjdbGTFtagExonParentTranscript)s''' % pdict
45 elif sjdbFileChrStartEnd:
46 cl += '--sjdbFileChrStartEnd %(sjdbFileChrStartEnd)s --sjdbOverhang %(sjdbOverhangs)s'
47 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-rnastar-index-builder-stderr" )
48 args = cl.split(' ')
49 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() )
50 return_code = proc.wait()
51 if return_code:
52 tmp_stderr.flush()
53 tmp_stderr.seek(0)
54 print >> sys.stderr, "Error building index: retcode=",retcode
55 while True:
56 chunk = tmp_stderr.read( CHUNK_SIZE )
57 if not chunk:
58 break
59 sys.stderr.write( chunk )
60 tmp_stderr.close()
61 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
62 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
63
64 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
65 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
66 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
67 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
68 return data_manager_dict
69
70 def main():
71 #Parse Command Line
72 parser = optparse.OptionParser()
73 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
74 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
75 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
76 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
77 parser.add_option( '--out_file', default=None)
78 parser.add_option( '--out_index_path', default=None)
79 parser.add_option( '--sjdbGTFfile', type="string", default=None )
80 parser.add_option( '--sjdbGTFchrPrefix', type="string", default=None )
81 parser.add_option( '--sjdbGTFfeatureExon', type="string", default=None )
82 parser.add_option( '--sjdbGTFtagExonParentTranscript', type="string", default=None )
83 parser.add_option( '--sjdbFileChrStartEnd', type="string", default=None )
84 parser.add_option( '--sjdbOverhang', type="int", default=100 )
85 parser.add_option( '--runThreadN', type="int", default=4 )
86 (options, args) = parser.parse_args()
87
88 filename = options.out_file
89 params = from_json_string( open( filename ).read() )
90 target_directory = options.out_index_path
91 os.mkdir( target_directory )
92 data_manager_dict = {}
93
94 dbkey = options.fasta_dbkey
95
96 if dbkey in [ None, '', '?' ]:
97 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
98
99 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
100
101 #build the index
102 build_rnastar_index( data_manager_dict, options.fasta_filename, target_directory, dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name,
103 sjdbOverhang=options.sjdbOverhang,sjdbGTFfile=options.sjdbGTFfile,
104 sjdbFileChrStartEnd=options.sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript=options.sjdbGTFtagExonParentTranscript,
105 sjdbGTFfeatureExon=options.sjdbGTFfeatureExon,sjdbGTFchrPrefix=options.sjdbGTFchrPrefix,
106 n_threads=options.runThreadN )
107
108 #save info to json file
109 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) )
110
111 if __name__ == "__main__": main()