Mercurial > repos > devteam > data_manager_bwa_index_builder
comparison data_manager/bwa_index_builder.py @ 0:ae10d0ddbbb1 draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_bwa_index_builder commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
| author | devteam |
|---|---|
| date | Fri, 06 Nov 2015 14:15:00 -0500 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:ae10d0ddbbb1 |
|---|---|
| 1 #!/usr/bin/env python | |
| 2 #Dan Blankenberg | |
| 3 | |
| 4 import sys | |
| 5 import os | |
| 6 import tempfile | |
| 7 import optparse | |
| 8 import subprocess | |
| 9 | |
| 10 from json import loads, dumps | |
| 11 | |
| 12 | |
| 13 CHUNK_SIZE = 2**20 | |
| 14 ONE_GB = 2**30 | |
| 15 | |
| 16 DEFAULT_DATA_TABLE_NAME = "bwa_indexes" | |
| 17 | |
| 18 def get_id_name( params, dbkey, fasta_description=None): | |
| 19 #TODO: ensure sequence_id is unique and does not already appear in location file | |
| 20 sequence_id = params['param_dict']['sequence_id'] | |
| 21 if not sequence_id: | |
| 22 sequence_id = dbkey | |
| 23 | |
| 24 sequence_name = params['param_dict']['sequence_name'] | |
| 25 if not sequence_name: | |
| 26 sequence_name = fasta_description | |
| 27 if not sequence_name: | |
| 28 sequence_name = dbkey | |
| 29 return sequence_id, sequence_name | |
| 30 | |
| 31 def build_bwa_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME, color_space = False ): | |
| 32 #TODO: allow multiple FASTA input files | |
| 33 #tmp_dir = tempfile.mkdtemp( prefix='tmp-data-manager-bwa-index-builder-' ) | |
| 34 fasta_base_name = os.path.split( fasta_filename )[-1] | |
| 35 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) | |
| 36 os.symlink( fasta_filename, sym_linked_fasta_filename ) | |
| 37 if params['param_dict']['index_algorithm'] == 'automatic': | |
| 38 if os.stat( fasta_filename ).st_size <= ONE_GB: #use 1 GB as cut off for memory vs. max of 2gb database size; this is somewhat arbitrary | |
| 39 index_algorithm = 'is' | |
| 40 else: | |
| 41 index_algorithm = 'bwtsw' | |
| 42 else: | |
| 43 index_algorithm = params['param_dict']['index_algorithm'] | |
| 44 | |
| 45 args = [ 'bwa', 'index', '-a', index_algorithm ] | |
| 46 if color_space: | |
| 47 args.append( '-c' ) | |
| 48 args.append( sym_linked_fasta_filename ) | |
| 49 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory ) | |
| 50 return_code = proc.wait() | |
| 51 if return_code: | |
| 52 print >> sys.stderr, "Error building index." | |
| 53 sys.exit( return_code ) | |
| 54 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) | |
| 55 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) | |
| 56 | |
| 57 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): | |
| 58 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) | |
| 59 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) | |
| 60 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | |
| 61 return data_manager_dict | |
| 62 | |
| 63 def main(): | |
| 64 #Parse Command Line | |
| 65 parser = optparse.OptionParser() | |
| 66 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) | |
| 67 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) | |
| 68 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | |
| 69 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' ) | |
| 70 parser.add_option( '-c', '--color_space', dest='color_space', action='store_true', default=False, help='color_space' ) | |
| 71 (options, args) = parser.parse_args() | |
| 72 | |
| 73 filename = args[0] | |
| 74 | |
| 75 params = loads( open( filename ).read() ) | |
| 76 target_directory = params[ 'output_data' ][0]['extra_files_path'] | |
| 77 os.mkdir( target_directory ) | |
| 78 data_manager_dict = {} | |
| 79 | |
| 80 dbkey = options.fasta_dbkey | |
| 81 | |
| 82 if dbkey in [ None, '', '?' ]: | |
| 83 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) | |
| 84 | |
| 85 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description ) | |
| 86 | |
| 87 #build the index | |
| 88 build_bwa_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 ) | |
| 89 | |
| 90 #save info to json file | |
| 91 open( filename, 'wb' ).write( dumps( data_manager_dict ) ) | |
| 92 | |
| 93 if __name__ == "__main__": main() |
