Mercurial > repos > iuc > data_manager_hisat2_index_builder
comparison data_manager/hisat2_index_builder.py @ 5:8eac26f44d29 draft
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_hisat2_index_builder commit 05c191c21eba39e75a8c8d6e6a00d8db63552695"
| author | iuc |
|---|---|
| date | Mon, 18 Nov 2019 17:22:30 -0500 |
| parents | d210e1f185bd |
| children | a04cebcd77f4 |
comparison
equal
deleted
inserted
replaced
| 4:d210e1f185bd | 5:8eac26f44d29 |
|---|---|
| 10 from json import dumps, loads | 10 from json import dumps, loads |
| 11 | 11 |
| 12 DEFAULT_DATA_TABLE_NAME = "hisat2_indexes" | 12 DEFAULT_DATA_TABLE_NAME = "hisat2_indexes" |
| 13 | 13 |
| 14 | 14 |
| 15 def get_id_name( params, dbkey, fasta_description=None): | 15 def get_id_name(params, dbkey, fasta_description=None): |
| 16 # TODO: ensure sequence_id is unique and does not already appear in location file | 16 # TODO: ensure sequence_id is unique and does not already appear in location file |
| 17 sequence_id = params['param_dict']['sequence_id'] | 17 sequence_id = params['param_dict']['sequence_id'] |
| 18 if not sequence_id: | 18 if not sequence_id: |
| 19 sequence_id = dbkey | 19 sequence_id = dbkey |
| 20 | 20 |
| 24 if not sequence_name: | 24 if not sequence_name: |
| 25 sequence_name = dbkey | 25 sequence_name = dbkey |
| 26 return sequence_id, sequence_name | 26 return sequence_id, sequence_name |
| 27 | 27 |
| 28 | 28 |
| 29 def build_hisat_index( data_manager_dict, options, params, sequence_id, sequence_name ): | 29 def build_hisat_index(data_manager_dict, options, params, sequence_id, sequence_name): |
| 30 data_table_name = options.data_table_name or DEFAULT_DATA_TABLE_NAME | 30 data_table_name = options.data_table_name or DEFAULT_DATA_TABLE_NAME |
| 31 target_directory = params[ 'output_data' ][0]['extra_files_path'] | 31 target_directory = params['output_data'][0]['extra_files_path'] |
| 32 if not os.path.exists( target_directory ): | 32 if not os.path.exists(target_directory): |
| 33 os.mkdir( target_directory ) | 33 os.mkdir(target_directory) |
| 34 fasta_base_name = os.path.split( options.fasta_filename )[-1] | 34 fasta_base_name = os.path.split(options.fasta_filename)[-1] |
| 35 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) | 35 sym_linked_fasta_filename = os.path.join(target_directory, fasta_base_name) |
| 36 os.symlink( options.fasta_filename, sym_linked_fasta_filename ) | 36 os.symlink(options.fasta_filename, sym_linked_fasta_filename) |
| 37 args = [ 'hisat2-build' ] | 37 args = ['hisat2-build'] |
| 38 args.extend( shlex.split( options.indexer_options ) ) | 38 args.extend(shlex.split(options.indexer_options)) |
| 39 args.extend( [ sym_linked_fasta_filename, sequence_id ] ) | 39 args.extend([sym_linked_fasta_filename, sequence_id]) |
| 40 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory ) | 40 proc = subprocess.Popen(args=args, shell=False, cwd=target_directory) |
| 41 return_code = proc.wait() | 41 return_code = proc.wait() |
| 42 if return_code: | 42 if return_code: |
| 43 print("Error building index.", file=sys.stderr) | 43 print("Error building index.", file=sys.stderr) |
| 44 sys.exit( return_code ) | 44 sys.exit(return_code) |
| 45 data_table_entry = dict( value=sequence_id, dbkey=options.fasta_dbkey, name=sequence_name, path=sequence_id ) | 45 data_table_entry = dict(value=sequence_id, dbkey=options.fasta_dbkey, name=sequence_name, path=sequence_id) |
| 46 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) | 46 _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry) |
| 47 | 47 |
| 48 | 48 |
| 49 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): | 49 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry): |
| 50 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) | 50 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
| 51 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) | 51 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, []) |
| 52 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | 52 data_manager_dict['data_tables'][data_table_name].append(data_table_entry) |
| 53 return data_manager_dict | 53 return data_manager_dict |
| 54 | 54 |
| 55 | 55 |
| 56 def main(): | 56 def main(): |
| 57 # Parse Command Line | 57 # Parse Command Line |
| 58 parser = argparse.ArgumentParser() | 58 parser = argparse.ArgumentParser() |
| 59 parser.add_argument( '--output', dest='output', action='store', type=str, default=None ) | 59 parser.add_argument('--output', dest='output', action='store', type=str, default=None) |
| 60 parser.add_argument( '--fasta_filename', dest='fasta_filename', action='store', type=str, default=None ) | 60 parser.add_argument('--fasta_filename', dest='fasta_filename', action='store', type=str, default=None) |
| 61 parser.add_argument( '--fasta_dbkey', dest='fasta_dbkey', action='store', type=str, default=None ) | 61 parser.add_argument('--fasta_dbkey', dest='fasta_dbkey', action='store', type=str, default=None) |
| 62 parser.add_argument( '--fasta_description', dest='fasta_description', action='store', type=str, default=None ) | 62 parser.add_argument('--fasta_description', dest='fasta_description', action='store', type=str, default=None) |
| 63 parser.add_argument( '--data_table_name', dest='data_table_name', action='store', type=str, default='hisat2_indexes' ) | 63 parser.add_argument('--data_table_name', dest='data_table_name', action='store', type=str, default='hisat2_indexes') |
| 64 parser.add_argument( '--indexer_options', dest='indexer_options', action='store', type=str, default='' ) | 64 parser.add_argument('--indexer_options', dest='indexer_options', action='store', type=str, default='') |
| 65 options = parser.parse_args() | 65 options = parser.parse_args() |
| 66 | 66 |
| 67 filename = options.output | 67 filename = options.output |
| 68 | 68 |
| 69 params = loads( open( filename ).read() ) | 69 params = loads(open(filename).read()) |
| 70 data_manager_dict = {} | 70 data_manager_dict = {} |
| 71 | 71 |
| 72 if options.fasta_dbkey in [ None, '', '?' ]: | 72 if options.fasta_dbkey in [None, '', '?']: |
| 73 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( options.fasta_dbkey ) ) | 73 raise Exception('"%s" is not a valid dbkey. You must specify a valid dbkey.' % (options.fasta_dbkey)) |
| 74 | 74 |
| 75 sequence_id, sequence_name = get_id_name( params, dbkey=options.fasta_dbkey, fasta_description=options.fasta_description ) | 75 sequence_id, sequence_name = get_id_name(params, dbkey=options.fasta_dbkey, fasta_description=options.fasta_description) |
| 76 | 76 |
| 77 # build the index | 77 # build the index |
| 78 build_hisat_index( data_manager_dict, options, params, sequence_id, sequence_name ) | 78 build_hisat_index(data_manager_dict, options, params, sequence_id, sequence_name) |
| 79 | 79 |
| 80 # save info to json file | 80 # save info to json file |
| 81 open( filename, 'w' ).write( dumps( data_manager_dict ) ) | 81 open(filename, 'w').write(dumps(data_manager_dict, sort_keys=True)) |
| 82 | 82 |
| 83 | 83 |
| 84 if __name__ == "__main__": | 84 if __name__ == "__main__": |
| 85 main() | 85 main() |
