Mercurial > repos > fubar > data_manager_rnasta
annotate 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 |
| 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 |
| 2 | 12 from json import loads, dumps |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
13 |
| 4 | 14 DEFAULT_DATA_TABLE_NAME = "rnastar_index" |
| 15 CHUNK_SIZE = 2**20 | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
16 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
17 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
|
18 #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
|
19 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
|
20 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
|
21 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
|
22 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
23 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
|
24 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
|
25 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
|
26 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
|
27 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
|
28 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
|
29 |
| 4 | 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, | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
50 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
|
51 #TODO: allow multiple FASTA input files |
| 4 | 52 fasta_base_name = os.path.basename( fasta_filename ) |
| 53 #sym_linked_fasta_filename = os.path.join( tmp_dir, fasta_base_name ) | |
| 54 #os.symlink( fasta_filename, sym_linked_fasta_filename ) | |
| 55 #print >> sys.stdout,'made',sym_linked_fasta_filename | |
| 56 cl = ['STAR','--runMode', 'genomeGenerate', '--genomeFastaFiles', fasta_filename, '--genomeDir', target_directory, '--runThreadN',n_threads ] | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
57 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
58 if sjdbGTFfile: |
| 4 | 59 cl += [ '--sjdbGTFfeatureExon', sjdbGTFfeatureExon,'--sjdbGTFtagExonParentTranscript',sjdbGTFtagExonParentTranscript] |
| 60 if (sjdbGTFchrPrefix > ''): | |
| 61 cl += ['--sjdbGTFchrPrefix', sjdbGTFchrPrefix] | |
| 62 cl += ['--sjdbOverhang', sjdbOverhang, '--sjdbGTFfile', sjdbGTFfile] | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
63 elif sjdbFileChrStartEnd: |
| 4 | 64 cl += ['--sjdbFileChrStartEnd', sjdbFileChrStartEnd, '--sjdbOverhang', sjdbOverhang] |
| 65 return_code = _run_command(command=cl,target_directory=target_directory) | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
66 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) |
| 4 | 67 data_manager_dict.setdefault('data_tables',{}) |
| 68 data_manager_dict['data_tables'].setdefault(data_table_name,[]) | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
69 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
|
70 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
|
71 |
| 4 | 72 |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
73 def main(): |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
74 #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
|
75 parser = optparse.OptionParser() |
| 4 | 76 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' ) | |
| 78 parser.add_option( '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | |
| 79 parser.add_option( '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' ) | |
|
0
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( '--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
|
81 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
|
82 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
|
83 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
|
84 parser.add_option( '--sjdbFileChrStartEnd', type="string", default=None ) |
| 4 | 85 parser.add_option( '--sjdbOverhang', type="string", default='100' ) |
| 86 parser.add_option( '--runThreadN', type="string", default='4' ) | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
87 (options, args) = parser.parse_args() |
| 4 | 88 filename = args[0] |
| 89 # this is passed as an positional parameter in the original wrapper with the extra_files_path from the params | |
| 2 | 90 params = loads( open( filename ).read() ) |
| 4 | 91 target_directory = params[ 'output_data' ][0]['extra_files_path'].encode('ascii','replace') |
| 92 try: | |
| 93 os.mkdir( target_directory ) | |
| 94 except OSError: | |
| 95 pass | |
| 96 | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
97 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
|
98 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
99 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
|
100 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
101 if dbkey in [ None, '', '?' ]: |
| 4 | 102 raise Exception( 'dbkey "%s" does not specify any locally known genome to be indexed.' % ( dbkey ) ) |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
103 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
104 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
|
105 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
106 #build the index |
| 4 | 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, | |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
109 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
|
110 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
|
111 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
|
112 n_threads=options.runThreadN ) |
| 1 | 113 |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
114 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
115 #save info to json file |
| 2 | 116 open( filename, 'wb' ).write( dumps( data_manager_dict ) ) |
|
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
117 |
|
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
118 if __name__ == "__main__": main() |
