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