Mercurial > repos > fubar > data_manager_rnasta

comparison data_manager/rnastar_index_builder.py @ 5:503097df1670 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author fubar
date Fri, 09 Jan 2015 22:38:21 -0500
parents 102bdfdda10b
children f9d8dc8c2152
comparison
equal deleted inserted replaced
4:102bdfdda10b 5:503097df1670
1 #!/usr/bin/env python 1 #!/usr/bin/env python
2 #Dan Blankenberg 2 #Dan Blankenberg
3 # adapted from Dan's BWA one for rna star 3 # adapted from Dan's BWA one for rna star
4 # ross lazarus sept 2014 4 # ross lazarus sept 2014
5 # 5 # fixed some stupid bugs January 2015
6
7
6 import sys 8 import sys
7 import os 9 import os
8 import tempfile 10 import tempfile
9 import optparse 11 import optparse
10 import subprocess 12 import subprocess
11 13
12 from json import loads, dumps 14 from galaxy.util.json import from_json_string, to_json_string
15
16
17 CHUNK_SIZE = 2**20
18 ONE_GB = 2**30
19
20 # from json import loads, dumps
13 21
14 DEFAULT_DATA_TABLE_NAME = "rnastar_index" 22 DEFAULT_DATA_TABLE_NAME = "rnastar_index"
15 CHUNK_SIZE = 2**20 23
16 24
17 def get_id_name( params, dbkey, fasta_description=None): 25 def get_id_name( params, dbkey, fasta_description=None):
18 #TODO: ensure sequence_id is unique and does not already appear in location file 26 #TODO: ensure sequence_id is unique and does not already appear in location file
19 sequence_id = params['param_dict']['sequence_id'] 27 sequence_id = params['param_dict']['sequence_id']
20 if not sequence_id: 28 if not sequence_id:
48 def build_rnastar_index( data_manager_dict, fasta_filename, target_directory, 56 def build_rnastar_index( data_manager_dict, fasta_filename, target_directory,
49 dbkey, sequence_id, sequence_name, data_table_name, 57 dbkey, sequence_id, sequence_name, data_table_name,
50 sjdbOverhang,sjdbGTFfile, sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript,sjdbGTFfeatureExon,sjdbGTFchrPrefix,n_threads): 58 sjdbOverhang,sjdbGTFfile, sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript,sjdbGTFfeatureExon,sjdbGTFchrPrefix,n_threads):
51 #TODO: allow multiple FASTA input files 59 #TODO: allow multiple FASTA input files
52 fasta_base_name = os.path.basename( fasta_filename ) 60 fasta_base_name = os.path.basename( fasta_filename )
53 #sym_linked_fasta_filename = os.path.join( tmp_dir, fasta_base_name ) 61 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
54 #os.symlink( fasta_filename, sym_linked_fasta_filename ) 62 os.symlink( fasta_filename, sym_linked_fasta_filename )
55 #print >> sys.stdout,'made',sym_linked_fasta_filename 63 print >> sys.stdout,'made',sym_linked_fasta_filename
56 cl = ['STAR','--runMode', 'genomeGenerate', '--genomeFastaFiles', fasta_filename, '--genomeDir', target_directory, '--runThreadN',n_threads ] 64 cl = ['STAR','--runMode', 'genomeGenerate', '--genomeFastaFiles', sym_linked_fasta_filename, '--genomeDir', target_directory, '--runThreadN',n_threads ]
57 65
58 if sjdbGTFfile: 66 if sjdbGTFfile:
59 cl += [ '--sjdbGTFfeatureExon', sjdbGTFfeatureExon,'--sjdbGTFtagExonParentTranscript',sjdbGTFtagExonParentTranscript] 67 cl += [ '--sjdbGTFfeatureExon', sjdbGTFfeatureExon,'--sjdbGTFtagExonParentTranscript',sjdbGTFtagExonParentTranscript]
60 if (sjdbGTFchrPrefix > ''): 68 if (sjdbGTFchrPrefix > ''):
61 cl += ['--sjdbGTFchrPrefix', sjdbGTFchrPrefix] 69 cl += ['--sjdbGTFchrPrefix', sjdbGTFchrPrefix]
62 cl += ['--sjdbOverhang', sjdbOverhang, '--sjdbGTFfile', sjdbGTFfile] 70 cl += ['--sjdbOverhang', sjdbOverhang, '--sjdbGTFfile', sjdbGTFfile]
63 elif sjdbFileChrStartEnd: 71 elif sjdbFileChrStartEnd:
64 cl += ['--sjdbFileChrStartEnd', sjdbFileChrStartEnd, '--sjdbOverhang', sjdbOverhang] 72 cl += ['--sjdbFileChrStartEnd', sjdbFileChrStartEnd, '--sjdbOverhang', sjdbOverhang]
65 return_code = _run_command(command=cl,target_directory=target_directory) 73
74 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-rnastar-index-builder-stderr" )
75 proc = subprocess.Popen( args=cl, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() )
76 return_code = proc.wait()
77 if return_code:
78 tmp_stderr.flush()
79 tmp_stderr.seek(0)
80 print >> sys.stderr, "Error building index:"
81 while True:
82 chunk = tmp_stderr.read( CHUNK_SIZE )
83 if not chunk:
84 break
85 sys.stderr.write( chunk )
86 sys.exit( return_code )
87 tmp_stderr.close()
66 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) 88 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
67 data_manager_dict.setdefault('data_tables',{}) 89 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
68 data_manager_dict['data_tables'].setdefault(data_table_name,[]) 90
91
92 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
93 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
94 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
69 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) 95 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
70 return data_manager_dict 96 return data_manager_dict
71 97
72
73 def main(): 98 def main():
74 #Parse Command Line 99 #Parse Command Line
75 parser = optparse.OptionParser() 100 parser = optparse.OptionParser()
76 parser.add_option( '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) 101 parser.add_option( '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
77 parser.add_option( '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) 102 parser.add_option( '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
83 parser.add_option( '--sjdbGTFtagExonParentTranscript', type="string", default=None ) 108 parser.add_option( '--sjdbGTFtagExonParentTranscript', type="string", default=None )
84 parser.add_option( '--sjdbFileChrStartEnd', type="string", default=None ) 109 parser.add_option( '--sjdbFileChrStartEnd', type="string", default=None )
85 parser.add_option( '--sjdbOverhang', type="string", default='100' ) 110 parser.add_option( '--sjdbOverhang', type="string", default='100' )
86 parser.add_option( '--runThreadN', type="string", default='4' ) 111 parser.add_option( '--runThreadN', type="string", default='4' )
87 (options, args) = parser.parse_args() 112 (options, args) = parser.parse_args()
88 filename = args[0] 113 filename = args[0]
89 # this is passed as an positional parameter in the original wrapper with the extra_files_path from the params 114 params = from_json_string( open( filename ).read() )
90 params = loads( open( filename ).read() )
91 target_directory = params[ 'output_data' ][0]['extra_files_path'].encode('ascii','replace') 115 target_directory = params[ 'output_data' ][0]['extra_files_path'].encode('ascii','replace')
116 data_manager_dict = {}
117 dbkey = options.fasta_dbkey
118 if dbkey in [ None, '', '?' ]:
119 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
120
121 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
122
92 try: 123 try:
93 os.mkdir( target_directory ) 124 os.mkdir( target_directory )
94 except OSError: 125 except OSError:
95 pass 126 pass
96
97 data_manager_dict = {}
98
99 dbkey = options.fasta_dbkey
100
101 if dbkey in [ None, '', '?' ]:
102 raise Exception( 'dbkey "%s" does not specify any locally known genome to be indexed.' % ( dbkey ) )
103
104 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
105
106 #build the index 127 #build the index
107 data_manager_dict = build_rnastar_index( data_manager_dict = data_manager_dict, fasta_filename = options.fasta_filename, target_directory = target_directory, 128 build_rnastar_index( data_manager_dict = data_manager_dict, fasta_filename = options.fasta_filename, target_directory = target_directory,
108 dbkey = dbkey, sequence_id = sequence_id, sequence_name = sequence_name, data_table_name=options.data_table_name, 129 dbkey = dbkey, sequence_id = sequence_id, sequence_name = sequence_name, data_table_name=options.data_table_name,
109 sjdbOverhang=options.sjdbOverhang,sjdbGTFfile=options.sjdbGTFfile, 130 sjdbOverhang=options.sjdbOverhang,sjdbGTFfile=options.sjdbGTFfile,
110 sjdbFileChrStartEnd=options.sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript=options.sjdbGTFtagExonParentTranscript, 131 sjdbFileChrStartEnd=options.sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript=options.sjdbGTFtagExonParentTranscript,
111 sjdbGTFfeatureExon=options.sjdbGTFfeatureExon,sjdbGTFchrPrefix=options.sjdbGTFchrPrefix, 132 sjdbGTFfeatureExon=options.sjdbGTFfeatureExon,sjdbGTFchrPrefix=options.sjdbGTFchrPrefix,
112 n_threads=options.runThreadN ) 133 n_threads=options.runThreadN )
113 134 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) )
114
115 #save info to json file
116 open( filename, 'wb' ).write( dumps( data_manager_dict ) )
117 135
118 if __name__ == "__main__": main() 136 if __name__ == "__main__": main()
137