comparison data_manager/rnastar_index_builder.py @ 4:102bdfdda10b draft

Uploaded
author fubar
date Thu, 08 Jan 2015 17:53:14 -0500
parents 8a2d16bfdae2
children 503097df1670
comparison
equal deleted inserted replaced
3:2468ef56337a 4:102bdfdda10b
9 import optparse 9 import optparse
10 import subprocess 10 import subprocess
11 11
12 from json import loads, dumps 12 from json import loads, dumps
13 13
14 DEFAULT_DATA_TABLE_NAME = "rnastar_indexes" 14 DEFAULT_DATA_TABLE_NAME = "rnastar_index"
15 CHUNK_SIZE = 2**20
15 16
16 def get_id_name( params, dbkey, fasta_description=None): 17 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 #TODO: ensure sequence_id is unique and does not already appear in location file
18 sequence_id = params['param_dict']['sequence_id'] 19 sequence_id = params['param_dict']['sequence_id']
19 if not sequence_id: 20 if not sequence_id:
24 sequence_name = fasta_description 25 sequence_name = fasta_description
25 if not sequence_name: 26 if not sequence_name:
26 sequence_name = dbkey 27 sequence_name = dbkey
27 return sequence_id, sequence_name 28 return sequence_id, sequence_name
28 29
29 def build_rnastar_index( data_manager_dict, fasta_filename, target_directory, dbkey, sequence_id, sequence_name, data_table_name, 30
31 def _run_command( command, target_directory ):
32 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-rnastar_index_builder-stderr" )
33 return_code = subprocess.call( command, shell=False, stdout=sys.stdout, stderr=tmp_stderr.fileno() )
34 if return_code:
35 tmp_stderr.flush()
36 tmp_stderr.seek( 0 )
37 sys.stderr.write( "### star reports an error:\n" )
38 while True:
39 chunk = tmp_stderr.read( CHUNK_SIZE )
40 if not chunk:
41 break
42 sys.stderr.write( chunk )
43 sys.exit( return_code )
44 tmp_stderr.close()
45 return return_code
46
47
48 def build_rnastar_index( data_manager_dict, fasta_filename, target_directory,
49 dbkey, sequence_id, sequence_name, data_table_name,
30 sjdbOverhang,sjdbGTFfile, sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript,sjdbGTFfeatureExon,sjdbGTFchrPrefix,n_threads): 50 sjdbOverhang,sjdbGTFfile, sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript,sjdbGTFfeatureExon,sjdbGTFchrPrefix,n_threads):
31 #TODO: allow multiple FASTA input files 51 #TODO: allow multiple FASTA input files
32 #tmp_dir = tempfile.mkdtemp( prefix='tmp-data-manager-bwa-index-builder-' ) 52 fasta_base_name = os.path.basename( fasta_filename )
33 fasta_base_name = os.path.split( fasta_filename )[-1] 53 #sym_linked_fasta_filename = os.path.join( tmp_dir, fasta_base_name )
34 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) 54 #os.symlink( fasta_filename, sym_linked_fasta_filename )
35 os.symlink( fasta_filename, sym_linked_fasta_filename ) 55 #print >> sys.stdout,'made',sym_linked_fasta_filename
36 pdict={'target_directory':target_directory,'n_threads':n_threads, 'sjdbFileChrStartEnd':sjdbFileChrStartEnd, 56 cl = ['STAR','--runMode', 'genomeGenerate', '--genomeFastaFiles', fasta_filename, '--genomeDir', target_directory, '--runThreadN',n_threads ]
37 'sjdbGTFtagExonParentTranscript':sjdbGTFtagExonParentTranscript, 'sjdbGTFfeatureExon':sjdbGTFfeatureExon,
38 'sjdbGTFchrPrefix':sjdbGTFchrPrefix,'sjdbOverhang':sjdbOverhang, 'sjdbGTFfile':sjdbGTFfile,
39 'sym_linked_fasta_filename':sym_linked_fasta_filename}
40 57
41 cl = 'STAR --runMode genomeGenerate --genomeFastaFiles %(sym_linked_fasta_filename)s --genomeDir %(target_directory)s --runThreadN %(n_threads)s' % pdict
42 if sjdbGTFfile: 58 if sjdbGTFfile:
43 cl += '''--sjdbGTFchrPrefix %(sjdbGTFchrPrefix)s --sjdbGTFfeatureExon %(sjdbGTFfeatureExon)s --sjdbOverhang %(sjdbOverhang)s 59 cl += [ '--sjdbGTFfeatureExon', sjdbGTFfeatureExon,'--sjdbGTFtagExonParentTranscript',sjdbGTFtagExonParentTranscript]
44 --sjdbGTFfile %(sjdbGTFfile)s --sjdbGTFtagExonParentTranscript %(sjdbGTFtagExonParentTranscript)s''' % pdict 60 if (sjdbGTFchrPrefix > ''):
61 cl += ['--sjdbGTFchrPrefix', sjdbGTFchrPrefix]
62 cl += ['--sjdbOverhang', sjdbOverhang, '--sjdbGTFfile', sjdbGTFfile]
45 elif sjdbFileChrStartEnd: 63 elif sjdbFileChrStartEnd:
46 cl += '--sjdbFileChrStartEnd %(sjdbFileChrStartEnd)s --sjdbOverhang %(sjdbOverhangs)s' % pdict 64 cl += ['--sjdbFileChrStartEnd', sjdbFileChrStartEnd, '--sjdbOverhang', sjdbOverhang]
47 args = cl.split(' ') 65 return_code = _run_command(command=cl,target_directory=target_directory)
48 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory )
49 return_code = proc.wait()
50 if return_code:
51 sys.exit( return_code )
52 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) 66 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
53 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) 67 data_manager_dict.setdefault('data_tables',{})
54 68 data_manager_dict['data_tables'].setdefault(data_table_name,[])
55 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
56 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
57 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
58 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) 69 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
59 return data_manager_dict 70 return data_manager_dict
71
60 72
61 def main(): 73 def main():
62 #Parse Command Line 74 #Parse Command Line
63 parser = optparse.OptionParser() 75 parser = optparse.OptionParser()
64 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) 76 parser.add_option( '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
65 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) 77 parser.add_option( '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
66 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) 78 parser.add_option( '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
67 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' ) 79 parser.add_option( '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
68 parser.add_option( '--out_file', default=None)
69 parser.add_option( '--out_index_path', default=None)
70 parser.add_option( '--sjdbGTFfile', type="string", default=None ) 80 parser.add_option( '--sjdbGTFfile', type="string", default=None )
71 parser.add_option( '--sjdbGTFchrPrefix', type="string", default=None ) 81 parser.add_option( '--sjdbGTFchrPrefix', type="string", default=None )
72 parser.add_option( '--sjdbGTFfeatureExon', type="string", default=None ) 82 parser.add_option( '--sjdbGTFfeatureExon', type="string", default=None )
73 parser.add_option( '--sjdbGTFtagExonParentTranscript', type="string", default=None ) 83 parser.add_option( '--sjdbGTFtagExonParentTranscript', type="string", default=None )
74 parser.add_option( '--sjdbFileChrStartEnd', type="string", default=None ) 84 parser.add_option( '--sjdbFileChrStartEnd', type="string", default=None )
75 parser.add_option( '--sjdbOverhang', type="int", default=100 ) 85 parser.add_option( '--sjdbOverhang', type="string", default='100' )
76 parser.add_option( '--runThreadN', type="int", default=4 ) 86 parser.add_option( '--runThreadN', type="string", default='4' )
77 (options, args) = parser.parse_args() 87 (options, args) = parser.parse_args()
78 88 filename = args[0]
79 filename = options.out_file 89 # this is passed as an positional parameter in the original wrapper with the extra_files_path from the params
80 params = loads( open( filename ).read() ) 90 params = loads( open( filename ).read() )
81 target_directory = options.out_index_path 91 target_directory = params[ 'output_data' ][0]['extra_files_path'].encode('ascii','replace')
82 os.mkdirs( target_directory ) 92 try:
93 os.mkdir( target_directory )
94 except OSError:
95 pass
96
83 data_manager_dict = {} 97 data_manager_dict = {}
84 98
85 dbkey = options.fasta_dbkey 99 dbkey = options.fasta_dbkey
86 100
87 if dbkey in [ None, '', '?' ]: 101 if dbkey in [ None, '', '?' ]:
88 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) 102 raise Exception( 'dbkey "%s" does not specify any locally known genome to be indexed.' % ( dbkey ) )
89 103
90 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 )
91 105
92 #build the index 106 #build the index
93 build_rnastar_index( data_manager_dict, options.fasta_filename, target_directory, dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name, 107 data_manager_dict = 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,
94 sjdbOverhang=options.sjdbOverhang,sjdbGTFfile=options.sjdbGTFfile, 109 sjdbOverhang=options.sjdbOverhang,sjdbGTFfile=options.sjdbGTFfile,
95 sjdbFileChrStartEnd=options.sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript=options.sjdbGTFtagExonParentTranscript, 110 sjdbFileChrStartEnd=options.sjdbFileChrStartEnd,sjdbGTFtagExonParentTranscript=options.sjdbGTFtagExonParentTranscript,
96 sjdbGTFfeatureExon=options.sjdbGTFfeatureExon,sjdbGTFchrPrefix=options.sjdbGTFchrPrefix, 111 sjdbGTFfeatureExon=options.sjdbGTFfeatureExon,sjdbGTFchrPrefix=options.sjdbGTFchrPrefix,
97 n_threads=options.runThreadN ) 112 n_threads=options.runThreadN )
98 113