Mercurial > repos > rnateam > data_manager_bowtie
comparison data_manager/bowtie_index_builder.py @ 0:07578a72b9b2 draft
Uploaded
| author | rnateam |
|---|---|
| date | Tue, 10 Feb 2015 13:35:43 -0500 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:07578a72b9b2 |
|---|---|
| 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 = "bowtie_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_bowtie_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME, color_space = False ): | |
| 30 #TODO: allow multiple FASTA input files | |
| 31 #tmp_dir = tempfile.mkdtemp( prefix='tmp-data-manager-bowtie-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 = [ 'bowtie-build' ] | |
| 36 if color_space: | |
| 37 args.append( '-c' ) | |
| 38 args.append( sym_linked_fasta_filename) | |
| 39 args.append( fasta_base_name ) | |
| 40 args.append( sym_linked_fasta_filename ) | |
| 41 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-bowtie-index-builder-stderr" ) | |
| 42 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() ) | |
| 43 return_code = proc.wait() | |
| 44 if return_code: | |
| 45 tmp_stderr.flush() | |
| 46 tmp_stderr.seek(0) | |
| 47 print >> sys.stderr, "Error building index:" | |
| 48 while True: | |
| 49 chunk = tmp_stderr.read( CHUNK_SIZE ) | |
| 50 if not chunk: | |
| 51 break | |
| 52 sys.stderr.write( chunk ) | |
| 53 sys.exit( return_code ) | |
| 54 tmp_stderr.close() | |
| 55 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) | |
| 56 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) | |
| 57 | |
| 58 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): | |
| 59 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) | |
| 60 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) | |
| 61 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | |
| 62 return data_manager_dict | |
| 63 | |
| 64 def main(): | |
| 65 #Parse Command Line | |
| 66 parser = optparse.OptionParser() | |
| 67 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) | |
| 68 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) | |
| 69 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | |
| 70 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' ) | |
| 71 parser.add_option( '-c', '--color_space', dest='color_space', action='store_true', default=False, help='color_space' ) | |
| 72 (options, args) = parser.parse_args() | |
| 73 | |
| 74 filename = args[0] | |
| 75 | |
| 76 params = from_json_string( open( filename ).read() ) | |
| 77 target_directory = params[ 'output_data' ][0]['extra_files_path'] | |
| 78 os.mkdir( target_directory ) | |
| 79 data_manager_dict = {} | |
| 80 | |
| 81 dbkey = options.fasta_dbkey | |
| 82 | |
| 83 if dbkey in [ None, '', '?' ]: | |
| 84 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) | |
| 85 | |
| 86 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description ) | |
| 87 | |
| 88 #build the index | |
| 89 build_bowtie_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, color_space=options.color_space ) | |
| 90 | |
| 91 #save info to json file | |
| 92 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) ) | |
| 93 | |
| 94 if __name__ == "__main__": main() |
