Mercurial > repos > fubar > data_manager_rnasta
annotate data_manager/rnastar_index_builder.py @ 5:503097df1670 draft
Uploaded
author | fubar |
---|---|
date | Fri, 09 Jan 2015 22:38:21 -0500 |
parents | 102bdfdda10b |
children | f9d8dc8c2152 |
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 |
5 | 5 # fixed some stupid bugs January 2015 |
6 | |
7 | |
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
8 import sys |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
9 import os |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
10 import tempfile |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
11 import optparse |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
12 import subprocess |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
13 |
5 | 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 | |
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
21 |
4 | 22 DEFAULT_DATA_TABLE_NAME = "rnastar_index" |
5 | 23 |
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
24 |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
25 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
|
26 #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
|
27 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
|
28 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
|
29 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
|
30 |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
31 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
|
32 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
|
33 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
|
34 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
|
35 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
|
36 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
|
37 |
4 | 38 |
39 def _run_command( command, target_directory ): | |
40 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-rnastar_index_builder-stderr" ) | |
41 return_code = subprocess.call( command, shell=False, stdout=sys.stdout, stderr=tmp_stderr.fileno() ) | |
42 if return_code: | |
43 tmp_stderr.flush() | |
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 | |
56 def build_rnastar_index( data_manager_dict, fasta_filename, target_directory, | |
57 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
|
58 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
|
59 #TODO: allow multiple FASTA input files |
4 | 60 fasta_base_name = os.path.basename( fasta_filename ) |
5 | 61 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) |
62 os.symlink( fasta_filename, sym_linked_fasta_filename ) | |
63 print >> sys.stdout,'made',sym_linked_fasta_filename | |
64 cl = ['STAR','--runMode', 'genomeGenerate', '--genomeFastaFiles', sym_linked_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
|
65 |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
66 if sjdbGTFfile: |
4 | 67 cl += [ '--sjdbGTFfeatureExon', sjdbGTFfeatureExon,'--sjdbGTFtagExonParentTranscript',sjdbGTFtagExonParentTranscript] |
68 if (sjdbGTFchrPrefix > ''): | |
69 cl += ['--sjdbGTFchrPrefix', sjdbGTFchrPrefix] | |
70 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
|
71 elif sjdbFileChrStartEnd: |
4 | 72 cl += ['--sjdbFileChrStartEnd', sjdbFileChrStartEnd, '--sjdbOverhang', sjdbOverhang] |
5 | 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() | |
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
88 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) |
5 | 89 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) |
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, [] ) | |
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
95 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
|
96 return data_manager_dict |
5 | 97 |
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
98 def main(): |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
99 #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
|
100 parser = optparse.OptionParser() |
4 | 101 parser.add_option( '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) |
102 parser.add_option( '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) | |
103 parser.add_option( '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | |
104 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
|
105 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
|
106 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
|
107 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
|
108 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
|
109 parser.add_option( '--sjdbFileChrStartEnd', type="string", default=None ) |
4 | 110 parser.add_option( '--sjdbOverhang', type="string", default='100' ) |
111 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
|
112 (options, args) = parser.parse_args() |
5 | 113 filename = args[0] |
114 params = from_json_string( open( filename ).read() ) | |
4 | 115 target_directory = params[ 'output_data' ][0]['extra_files_path'].encode('ascii','replace') |
5 | 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 | |
4 | 123 try: |
124 os.mkdir( target_directory ) | |
125 except OSError: | |
126 pass | |
0
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
127 #build the index |
5 | 128 build_rnastar_index( data_manager_dict = data_manager_dict, fasta_filename = options.fasta_filename, target_directory = target_directory, |
4 | 129 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
|
130 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
|
131 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
|
132 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
|
133 n_threads=options.runThreadN ) |
5 | 134 open( filename, 'wb' ).write( to_json_string( 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
|
135 |
84f6e67cbae5
Initial commit - problems testing on my ram starved laptop - need to test on a bigger ram machine
fubar
parents:
diff
changeset
|
136 if __name__ == "__main__": main() |
5 | 137 |