Mercurial > repos > jjohnson > data_manager_hisat_index_builder
comparison data_manager/hisat_index_builder.py @ 0:4c7db9cbbbf7 draft default tip
Uploaded
author | jjohnson |
---|---|
date | Tue, 22 Jul 2014 08:52:53 -0400 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:4c7db9cbbbf7 |
---|---|
1 #!/usr/bin/env python | |
2 | |
3 import sys | |
4 import os | |
5 import tempfile | |
6 import optparse | |
7 import subprocess | |
8 | |
9 from galaxy.util.json import from_json_string, to_json_string | |
10 | |
11 | |
12 CHUNK_SIZE = 2**20 | |
13 | |
14 DEFAULT_DATA_TABLE_NAME = "hisat_indexes" | |
15 | |
16 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 sequence_id = params['param_dict']['sequence_id'] | |
19 if not sequence_id: | |
20 sequence_id = dbkey | |
21 | |
22 sequence_name = params['param_dict']['sequence_name'] | |
23 if not sequence_name: | |
24 sequence_name = fasta_description | |
25 if not sequence_name: | |
26 sequence_name = dbkey | |
27 return sequence_id, sequence_name | |
28 | |
29 def build_hisat_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME): | |
30 #TODO: allow multiple FASTA input files | |
31 #tmp_dir = tempfile.mkdtemp( prefix='tmp-data-manager-hisat-index-builder-' ) | |
32 fasta_base_name = os.path.split( fasta_filename )[-1] | |
33 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) | |
34 os.symlink( fasta_filename, sym_linked_fasta_filename ) | |
35 args = [ 'hisat-build',sym_linked_fasta_filename,fasta_base_name ] | |
36 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-hisat-index-builder-stderr" ) | |
37 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() ) | |
38 return_code = proc.wait() | |
39 if return_code: | |
40 tmp_stderr.flush() | |
41 tmp_stderr.seek(0) | |
42 print >> sys.stderr, "Error building index:" | |
43 while True: | |
44 chunk = tmp_stderr.read( CHUNK_SIZE ) | |
45 if not chunk: | |
46 break | |
47 sys.stderr.write( chunk ) | |
48 sys.exit( return_code ) | |
49 tmp_stderr.close() | |
50 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) | |
51 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) | |
52 | |
53 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): | |
54 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) | |
55 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) | |
56 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | |
57 return data_manager_dict | |
58 | |
59 def main(): | |
60 #Parse Command Line | |
61 parser = optparse.OptionParser() | |
62 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) | |
63 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) | |
64 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | |
65 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' ) | |
66 (options, args) = parser.parse_args() | |
67 | |
68 filename = args[0] | |
69 | |
70 params = from_json_string( open( filename ).read() ) | |
71 target_directory = params[ 'output_data' ][0]['extra_files_path'] | |
72 os.mkdir( target_directory ) | |
73 data_manager_dict = {} | |
74 | |
75 dbkey = options.fasta_dbkey | |
76 | |
77 if dbkey in [ None, '', '?' ]: | |
78 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) | |
79 | |
80 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description ) | |
81 | |
82 #build the index | |
83 build_hisat_index( data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME ) | |
84 | |
85 #save info to json file | |
86 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) ) | |
87 | |
88 if __name__ == "__main__": main() |