Mercurial > repos > devteam > data_manager_twobit_builder
comparison data_manager/twobit_builder.py @ 3:5c77d4091880 draft
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_twobit_builder commit 82ad45497a7b2277a710c634378a4122fb75ce1f"
| author | iuc |
|---|---|
| date | Mon, 18 Nov 2019 17:43:27 -0500 |
| parents | 9946bc39c834 |
| children | 625fd9476402 |
comparison
equal
deleted
inserted
replaced
| 2:9946bc39c834 | 3:5c77d4091880 |
|---|---|
| 10 from json import dumps, loads | 10 from json import dumps, loads |
| 11 | 11 |
| 12 CHUNK_SIZE = 2**20 # 1mb | 12 CHUNK_SIZE = 2**20 # 1mb |
| 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 # uuid.uuid4() generate and use an uuid | 19 sequence_id = dbkey # uuid.uuid4() generate and use an uuid |
| 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_twobit( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name ): | 29 def build_twobit(data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name): |
| 30 twobit_base_name = "%s.2bit" % ( sequence_id ) | 30 twobit_base_name = "%s.2bit" % (sequence_id) |
| 31 twobit_filename = os.path.join( target_directory, twobit_base_name ) | 31 twobit_filename = os.path.join(target_directory, twobit_base_name) |
| 32 | 32 |
| 33 args = [ 'faToTwoBit', fasta_filename, twobit_filename ] | 33 args = ['faToTwoBit', fasta_filename, twobit_filename] |
| 34 tmp_stderr = tempfile.NamedTemporaryFile( prefix="tmp-data-manager-twobit-builder-stderr" ) | 34 tmp_stderr = tempfile.NamedTemporaryFile(prefix="tmp-data-manager-twobit-builder-stderr") |
| 35 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() ) | 35 proc = subprocess.Popen(args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno()) |
| 36 return_code = proc.wait() | 36 return_code = proc.wait() |
| 37 if return_code: | 37 if return_code: |
| 38 tmp_stderr.flush() | 38 tmp_stderr.flush() |
| 39 tmp_stderr.seek(0) | 39 tmp_stderr.seek(0) |
| 40 print("Error building index:", file=sys.stderr) | 40 print("Error building index:", file=sys.stderr) |
| 41 while True: | 41 while True: |
| 42 chunk = tmp_stderr.read( CHUNK_SIZE ) | 42 chunk = tmp_stderr.read(CHUNK_SIZE) |
| 43 if not chunk: | 43 if not chunk: |
| 44 break | 44 break |
| 45 sys.stderr.write( chunk ) | 45 sys.stderr.write(chunk) |
| 46 sys.exit( return_code ) | 46 sys.exit(return_code) |
| 47 tmp_stderr.close() | 47 tmp_stderr.close() |
| 48 # lastz_seqs | 48 # lastz_seqs |
| 49 data_table_entry = dict( value=sequence_id, name=sequence_name, path=twobit_base_name ) | 49 data_table_entry = dict(value=sequence_id, name=sequence_name, path=twobit_base_name) |
| 50 | 50 |
| 51 _add_data_table_entry( data_manager_dict, "lastz_seqs", data_table_entry ) | 51 _add_data_table_entry(data_manager_dict, "lastz_seqs", data_table_entry) |
| 52 # twobit.loc | 52 # twobit.loc |
| 53 data_table_entry = dict( value=sequence_id, path=twobit_base_name ) | 53 data_table_entry = dict(value=sequence_id, path=twobit_base_name) |
| 54 | 54 |
| 55 _add_data_table_entry( data_manager_dict, "twobit", data_table_entry ) | 55 _add_data_table_entry(data_manager_dict, "twobit", data_table_entry) |
| 56 # alignseq | 56 # alignseq |
| 57 data_table_entry = dict( type="seq", value=sequence_id, path=twobit_base_name ) | 57 data_table_entry = dict(type="seq", value=sequence_id, path=twobit_base_name) |
| 58 | 58 |
| 59 _add_data_table_entry( data_manager_dict, "alignseq_seq", data_table_entry ) | 59 _add_data_table_entry(data_manager_dict, "alignseq_seq", data_table_entry) |
| 60 | 60 |
| 61 | 61 |
| 62 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): | 62 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry): |
| 63 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) | 63 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
| 64 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) | 64 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, []) |
| 65 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | 65 data_manager_dict['data_tables'][data_table_name].append(data_table_entry) |
| 66 return data_manager_dict | 66 return data_manager_dict |
| 67 | 67 |
| 68 | 68 |
| 69 def main(): | 69 def main(): |
| 70 parser = optparse.OptionParser() | 70 parser = optparse.OptionParser() |
| 71 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) | 71 parser.add_option('-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename') |
| 72 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) | 72 parser.add_option('-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey') |
| 73 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | 73 parser.add_option('-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description') |
| 74 (options, args) = parser.parse_args() | 74 (options, args) = parser.parse_args() |
| 75 | 75 |
| 76 filename = args[0] | 76 filename = args[0] |
| 77 | 77 |
| 78 params = loads( open( filename ).read() ) | 78 params = loads(open(filename).read()) |
| 79 | 79 |
| 80 target_directory = params[ 'output_data' ][0]['extra_files_path'] | 80 target_directory = params['output_data'][0]['extra_files_path'] |
| 81 os.mkdir( target_directory ) | 81 os.mkdir(target_directory) |
| 82 data_manager_dict = {} | 82 data_manager_dict = {} |
| 83 | 83 |
| 84 dbkey = options.fasta_dbkey | 84 dbkey = options.fasta_dbkey |
| 85 | 85 |
| 86 if dbkey in [ None, '', '?' ]: | 86 if dbkey in [None, '', '?']: |
| 87 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) | 87 raise Exception('"%s" is not a valid dbkey. You must specify a valid dbkey.' % (dbkey)) |
| 88 | 88 |
| 89 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description ) | 89 sequence_id, sequence_name = get_id_name(params, dbkey=dbkey, fasta_description=options.fasta_description) |
| 90 | 90 |
| 91 # build the index | 91 # build the index |
| 92 build_twobit( data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name ) | 92 build_twobit(data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name) |
| 93 | 93 |
| 94 # save info to json file | 94 # save info to json file |
| 95 open( filename, 'wb' ).write( dumps( data_manager_dict ) ) | 95 open(filename, 'w').write(dumps(data_manager_dict, sort_keys=True)) |
| 96 | 96 |
| 97 | 97 |
| 98 if __name__ == "__main__": | 98 if __name__ == "__main__": |
| 99 main() | 99 main() |
