Mercurial > repos > fubar > data_manager_rnasta
comparison data_manager/rnastar_index_builder.py @ 6:f9d8dc8c2152 draft
Uploaded
author | fubar |
---|---|
date | Fri, 09 Jan 2015 23:16:54 -0500 |
parents | 503097df1670 |
children | 5c7efaab6713 |
comparison
equal
deleted
inserted
replaced
5:503097df1670 | 6:f9d8dc8c2152 |
---|---|
34 if not sequence_name: | 34 if not sequence_name: |
35 sequence_name = dbkey | 35 sequence_name = dbkey |
36 return sequence_id, sequence_name | 36 return sequence_id, sequence_name |
37 | 37 |
38 | 38 |
39 def _run_command( command, target_directory ): | 39 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): |
40 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-rnastar_index_builder-stderr" ) | 40 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) |
41 return_code = subprocess.call( command, shell=False, stdout=sys.stdout, stderr=tmp_stderr.fileno() ) | 41 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) |
42 if return_code: | 42 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) |
43 tmp_stderr.flush() | 43 return data_manager_dict |
44 tmp_stderr.seek( 0 ) | |
45 sys.stderr.write( "### star reports an error:\n" ) | |
46 while True: | |
47 chunk = tmp_stderr.read( CHUNK_SIZE ) | |
48 if not chunk: | |
49 break | |
50 sys.stderr.write( chunk ) | |
51 sys.exit( return_code ) | |
52 tmp_stderr.close() | |
53 return return_code | |
54 | |
55 | 44 |
56 def build_rnastar_index( data_manager_dict, fasta_filename, target_directory, | 45 def build_rnastar_index( data_manager_dict, fasta_filename, target_directory, |
57 dbkey, sequence_id, sequence_name, data_table_name, | 46 dbkey, sequence_id, sequence_name, data_table_name, |
58 sjdbOverhang,sjdbGTFfile, sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript,sjdbGTFfeatureExon,sjdbGTFchrPrefix,n_threads): | 47 sjdbOverhang,sjdbGTFfile, sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript,sjdbGTFfeatureExon,sjdbGTFchrPrefix,n_threads): |
59 #TODO: allow multiple FASTA input files | 48 #TODO: allow multiple FASTA input files |
60 fasta_base_name = os.path.basename( fasta_filename ) | 49 fasta_base_name = os.path.basename( fasta_filename ) |
61 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) | 50 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) |
62 os.symlink( fasta_filename, sym_linked_fasta_filename ) | 51 os.symlink( fasta_filename, sym_linked_fasta_filename ) |
63 print >> sys.stdout,'made',sym_linked_fasta_filename | 52 # print >> sys.stdout,'made',sym_linked_fasta_filename |
64 cl = ['STAR','--runMode', 'genomeGenerate', '--genomeFastaFiles', sym_linked_fasta_filename, '--genomeDir', target_directory, '--runThreadN',n_threads ] | 53 cl = ['STAR','--runMode', 'genomeGenerate', '--genomeFastaFiles', sym_linked_fasta_filename, '--genomeDir', target_directory, '--runThreadN',n_threads ] |
65 | 54 |
66 if sjdbGTFfile: | 55 if sjdbGTFfile: |
67 cl += [ '--sjdbGTFfeatureExon', sjdbGTFfeatureExon,'--sjdbGTFtagExonParentTranscript',sjdbGTFtagExonParentTranscript] | 56 cl += [ '--sjdbGTFfeatureExon', sjdbGTFfeatureExon,'--sjdbGTFtagExonParentTranscript',sjdbGTFtagExonParentTranscript] |
68 if (sjdbGTFchrPrefix > ''): | 57 if (sjdbGTFchrPrefix > ''): |
84 break | 73 break |
85 sys.stderr.write( chunk ) | 74 sys.stderr.write( chunk ) |
86 sys.exit( return_code ) | 75 sys.exit( return_code ) |
87 tmp_stderr.close() | 76 tmp_stderr.close() |
88 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) | 77 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) |
89 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) | 78 data_manager_dict = _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) |
79 return data_manager_dict | |
90 | 80 |
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, [] ) | |
95 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | |
96 return data_manager_dict | |
97 | 81 |
98 def main(): | 82 def main(): |
99 #Parse Command Line | 83 #Parse Command Line |
100 parser = optparse.OptionParser() | 84 parser = optparse.OptionParser() |
101 parser.add_option( '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) | 85 parser.add_option( '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) |
111 parser.add_option( '--runThreadN', type="string", default='4' ) | 95 parser.add_option( '--runThreadN', type="string", default='4' ) |
112 (options, args) = parser.parse_args() | 96 (options, args) = parser.parse_args() |
113 filename = args[0] | 97 filename = args[0] |
114 params = from_json_string( open( filename ).read() ) | 98 params = from_json_string( open( filename ).read() ) |
115 target_directory = params[ 'output_data' ][0]['extra_files_path'].encode('ascii','replace') | 99 target_directory = params[ 'output_data' ][0]['extra_files_path'].encode('ascii','replace') |
116 data_manager_dict = {} | |
117 dbkey = options.fasta_dbkey | 100 dbkey = options.fasta_dbkey |
118 if dbkey in [ None, '', '?' ]: | 101 if dbkey in [ None, '', '?' ]: |
119 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) | 102 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) |
120 | 103 |
121 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description ) | 104 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description ) |
123 try: | 106 try: |
124 os.mkdir( target_directory ) | 107 os.mkdir( target_directory ) |
125 except OSError: | 108 except OSError: |
126 pass | 109 pass |
127 #build the index | 110 #build the index |
128 build_rnastar_index( data_manager_dict = data_manager_dict, fasta_filename = options.fasta_filename, target_directory = target_directory, | 111 data_manager_dict = build_rnastar_index( data_manager_dict = data_manager_dict, fasta_filename = options.fasta_filename, target_directory = target_directory, |
129 dbkey = dbkey, sequence_id = sequence_id, sequence_name = sequence_name, data_table_name=options.data_table_name, | 112 dbkey = dbkey, sequence_id = sequence_id, sequence_name = sequence_name, data_table_name=options.data_table_name, |
130 sjdbOverhang=options.sjdbOverhang,sjdbGTFfile=options.sjdbGTFfile, | 113 sjdbOverhang=options.sjdbOverhang,sjdbGTFfile=options.sjdbGTFfile, |
131 sjdbFileChrStartEnd=options.sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript=options.sjdbGTFtagExonParentTranscript, | 114 sjdbFileChrStartEnd=options.sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript=options.sjdbGTFtagExonParentTranscript, |
132 sjdbGTFfeatureExon=options.sjdbGTFfeatureExon,sjdbGTFchrPrefix=options.sjdbGTFchrPrefix, | 115 sjdbGTFfeatureExon=options.sjdbGTFfeatureExon,sjdbGTFchrPrefix=options.sjdbGTFchrPrefix, |
133 n_threads=options.runThreadN ) | 116 n_threads=options.runThreadN ) |