Mercurial > repos > devteam > data_manager_bowtie2_index_builder
comparison data_manager/bowtie2_index_builder.py @ 7:1c0f8e9d87c6 draft
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bowtie2_index_builder commit 9fa01fe96452d50646e066066b1e556ba4a75ced"
author | iuc |
---|---|
date | Mon, 06 Apr 2020 19:27:05 -0400 |
parents | d1d5feb31804 |
children | 1586b111d2a3 |
comparison
equal
deleted
inserted
replaced
6:847c9ef56a52 | 7:1c0f8e9d87c6 |
---|---|
9 from json import dumps, loads | 9 from json import dumps, loads |
10 | 10 |
11 DEFAULT_DATA_TABLE_NAMES = ["bowtie2_indexes"] | 11 DEFAULT_DATA_TABLE_NAMES = ["bowtie2_indexes"] |
12 | 12 |
13 | 13 |
14 def get_id_name( params, dbkey, fasta_description=None): | 14 def get_id_name(params, dbkey, fasta_description=None): |
15 # TODO: ensure sequence_id is unique and does not already appear in location file | 15 # TODO: ensure sequence_id is unique and does not already appear in location file |
16 sequence_id = params['param_dict']['sequence_id'] | 16 sequence_id = params['param_dict']['sequence_id'] |
17 if not sequence_id: | 17 if not sequence_id: |
18 sequence_id = dbkey | 18 sequence_id = dbkey |
19 | 19 |
23 if not sequence_name: | 23 if not sequence_name: |
24 sequence_name = dbkey | 24 sequence_name = dbkey |
25 return sequence_id, sequence_name | 25 return sequence_id, sequence_name |
26 | 26 |
27 | 27 |
28 def build_bowtie2_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_names=DEFAULT_DATA_TABLE_NAMES ): | 28 def build_bowtie2_index(data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_names=DEFAULT_DATA_TABLE_NAMES): |
29 # TODO: allow multiple FASTA input files | 29 # TODO: allow multiple FASTA input files |
30 fasta_base_name = os.path.split( fasta_filename )[-1] | 30 fasta_base_name = os.path.split(fasta_filename)[-1] |
31 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) | 31 sym_linked_fasta_filename = os.path.join(target_directory, fasta_base_name) |
32 os.symlink( fasta_filename, sym_linked_fasta_filename ) | 32 os.symlink(fasta_filename, sym_linked_fasta_filename) |
33 args = [ 'bowtie2-build', sym_linked_fasta_filename, sequence_id ] | 33 args = ['bowtie2-build', sym_linked_fasta_filename, sequence_id] |
34 threads = os.environ.get('GALAXY_SLOTS') | 34 threads = os.environ.get('GALAXY_SLOTS') |
35 if threads: | 35 if threads: |
36 args.extend(['--threads', threads]) | 36 args.extend(['--threads', threads]) |
37 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory ) | 37 proc = subprocess.Popen(args=args, shell=False, cwd=target_directory) |
38 return_code = proc.wait() | 38 return_code = proc.wait() |
39 if return_code: | 39 if return_code: |
40 print("Error building index.", file=sys.stderr) | 40 print("Error building index.", file=sys.stderr) |
41 sys.exit( return_code ) | 41 sys.exit(return_code) |
42 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=sequence_id ) | 42 data_table_entry = dict(value=sequence_id, dbkey=dbkey, name=sequence_name, path=sequence_id) |
43 for data_table_name in data_table_names: | 43 for data_table_name in data_table_names: |
44 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) | 44 _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry) |
45 | 45 |
46 | 46 |
47 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): | 47 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry): |
48 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) | 48 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
49 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) | 49 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, []) |
50 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | 50 data_manager_dict['data_tables'][data_table_name].append(data_table_entry) |
51 return data_manager_dict | 51 return data_manager_dict |
52 | 52 |
53 | 53 |
54 def main(): | 54 def main(): |
55 parser = optparse.OptionParser() | 55 parser = optparse.OptionParser() |
56 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) | 56 parser.add_option('-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename') |
57 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) | 57 parser.add_option('-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey') |
58 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | 58 parser.add_option('-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description') |
59 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='append', type="string", default=None, help='data_table_name' ) | 59 parser.add_option('-n', '--data_table_name', dest='data_table_name', action='append', type="string", default=None, help='data_table_name') |
60 (options, args) = parser.parse_args() | 60 (options, args) = parser.parse_args() |
61 | 61 |
62 filename = args[0] | 62 filename = args[0] |
63 | 63 |
64 params = loads( open( filename ).read() ) | 64 params = loads(open(filename).read()) |
65 target_directory = params[ 'output_data' ][0]['extra_files_path'] | 65 target_directory = params['output_data'][0]['extra_files_path'] |
66 os.mkdir( target_directory ) | 66 os.mkdir(target_directory) |
67 data_manager_dict = {} | 67 data_manager_dict = {} |
68 | 68 |
69 dbkey = options.fasta_dbkey | 69 dbkey = options.fasta_dbkey |
70 | 70 |
71 if dbkey in [ None, '', '?' ]: | 71 if dbkey in [None, '', '?']: |
72 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) | 72 raise Exception('"%s" is not a valid dbkey. You must specify a valid dbkey.' % (dbkey)) |
73 | 73 |
74 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description ) | 74 sequence_id, sequence_name = get_id_name(params, dbkey=dbkey, fasta_description=options.fasta_description) |
75 | 75 |
76 # build the index | 76 # build the index |
77 build_bowtie2_index( data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_names=options.data_table_name or DEFAULT_DATA_TABLE_NAMES ) | 77 build_bowtie2_index(data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_names=options.data_table_name or DEFAULT_DATA_TABLE_NAMES) |
78 | 78 |
79 # save info to json file | 79 # save info to json file |
80 with open(filename, 'w') as json_out: | 80 with open(filename, 'w') as json_out: |
81 json_out.write( dumps( data_manager_dict, sort_keys=True ) ) | 81 json_out.write(dumps(data_manager_dict, sort_keys=True)) |
82 | 82 |
83 | 83 |
84 if __name__ == "__main__": | 84 if __name__ == "__main__": |
85 main() | 85 main() |