Mercurial > repos > devteam > data_manager_sam_fasta_index_builder
comparison data_manager/data_manager_sam_fasta_index_builder.py @ 3:1f54e98616af draft
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_sam_fasta_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
author | iuc |
---|---|
date | Tue, 04 Apr 2017 18:13:56 -0400 |
parents | 2a1ac1abc3f7 |
children | cc4ef4d38cf9 |
comparison
equal
deleted
inserted
replaced
2:2a1ac1abc3f7 | 3:1f54e98616af |
---|---|
1 #!/usr/bin/env python | 1 #!/usr/bin/env python |
2 #Dan Blankenberg | 2 # Dan Blankenberg |
3 | 3 |
4 import json | 4 import json |
5 import optparse | 5 import optparse |
6 import os | 6 import os |
7 import subprocess | 7 import subprocess |
10 | 10 |
11 CHUNK_SIZE = 2**20 | 11 CHUNK_SIZE = 2**20 |
12 | 12 |
13 DEFAULT_DATA_TABLE_NAME = "fasta_indexes" | 13 DEFAULT_DATA_TABLE_NAME = "fasta_indexes" |
14 | 14 |
15 | |
15 def get_id_name( params, dbkey, fasta_description=None): | 16 def get_id_name( params, dbkey, fasta_description=None): |
16 #TODO: ensure sequence_id is unique and does not already appear in location file | 17 # TODO: ensure sequence_id is unique and does not already appear in location file |
17 sequence_id = params['param_dict']['sequence_id'] | 18 sequence_id = params['param_dict']['sequence_id'] |
18 if not sequence_id: | 19 if not sequence_id: |
19 sequence_id = dbkey | 20 sequence_id = dbkey |
20 | 21 |
21 sequence_name = params['param_dict']['sequence_name'] | 22 sequence_name = params['param_dict']['sequence_name'] |
22 if not sequence_name: | 23 if not sequence_name: |
23 sequence_name = fasta_description | 24 sequence_name = fasta_description |
24 if not sequence_name: | 25 if not sequence_name: |
25 sequence_name = dbkey | 26 sequence_name = dbkey |
26 return sequence_id, sequence_name | 27 return sequence_id, sequence_name |
27 | 28 |
29 | |
28 def build_sam_index( data_manager_dict, fasta_filename, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME ): | 30 def build_sam_index( data_manager_dict, fasta_filename, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME ): |
29 #TODO: allow multiple FASTA input files | 31 # TODO: allow multiple FASTA input files |
30 assert os.path.exists( fasta_filename ), 'FASTA file "%s" is missing, cannot build samtools index.' % fasta_filename | 32 assert os.path.exists( fasta_filename ), 'FASTA file "%s" is missing, cannot build samtools index.' % fasta_filename |
31 fasta_base_name = os.path.split( fasta_filename )[-1] | 33 fasta_base_name = os.path.split( fasta_filename )[-1] |
32 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) | 34 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) |
33 os.symlink( fasta_filename, sym_linked_fasta_filename ) | 35 os.symlink( fasta_filename, sym_linked_fasta_filename ) |
34 | 36 |
35 args = [ 'samtools', 'faidx' ] | 37 args = [ 'samtools', 'faidx' ] |
36 args.append( sym_linked_fasta_filename ) | 38 args.append( sym_linked_fasta_filename ) |
37 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-sam_fa_index_builder-stderr" ) | 39 tmp_stderr = tempfile.NamedTemporaryFile(prefix="tmp-data-manager-sam_fa_index_builder-stderr") |
38 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() ) | 40 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() ) |
39 return_code = proc.wait() | 41 return_code = proc.wait() |
40 if return_code: | 42 if return_code: |
41 tmp_stderr.flush() | 43 tmp_stderr.flush() |
42 tmp_stderr.seek( 0 ) | 44 tmp_stderr.seek( 0 ) |
49 sys.exit( return_code ) | 51 sys.exit( return_code ) |
50 tmp_stderr.close() | 52 tmp_stderr.close() |
51 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) | 53 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) |
52 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) | 54 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) |
53 | 55 |
56 | |
54 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): | 57 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): |
55 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) | 58 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) |
56 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) | 59 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) |
57 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | 60 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) |
58 return data_manager_dict | 61 return data_manager_dict |
59 | 62 |
63 | |
60 def main(): | 64 def main(): |
61 #Parse Command Line | |
62 parser = optparse.OptionParser() | 65 parser = optparse.OptionParser() |
63 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) | 66 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) |
64 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) | 67 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) |
65 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | 68 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) |
66 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' ) | 69 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' ) |
67 (options, args) = parser.parse_args() | 70 (options, args) = parser.parse_args() |
68 | 71 |
69 filename = args[0] | 72 filename = args[0] |
70 | 73 |
71 params = json.loads( open( filename ).read() ) | 74 params = json.loads( open( filename ).read() ) |
72 target_directory = params[ 'output_data' ][0]['extra_files_path'] | 75 target_directory = params[ 'output_data' ][0]['extra_files_path'] |
73 os.mkdir( target_directory ) | 76 os.mkdir( target_directory ) |
74 data_manager_dict = {} | 77 data_manager_dict = {} |
75 | 78 |
76 if options.fasta_dbkey in [ None, '', '?' ]: | 79 if options.fasta_dbkey in [ None, '', '?' ]: |
77 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( options.fasta_dbkey ) ) | 80 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( options.fasta_dbkey ) ) |
78 | 81 |
79 sequence_id, sequence_name = get_id_name( params, dbkey=options.fasta_dbkey, fasta_description=options.fasta_description ) | 82 sequence_id, sequence_name = get_id_name( params, dbkey=options.fasta_dbkey, fasta_description=options.fasta_description ) |
80 | 83 |
81 #build the index | 84 # build the index |
82 build_sam_index( data_manager_dict, options.fasta_filename, target_directory, options.fasta_dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME ) | 85 build_sam_index( data_manager_dict, options.fasta_filename, target_directory, options.fasta_dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME ) |
83 | 86 |
84 #save info to json file | 87 # save info to json file |
85 open( filename, 'wb' ).write( json.dumps( data_manager_dict ) ) | 88 open( filename, 'wb' ).write( json.dumps( data_manager_dict ) ) |
86 | 89 |
87 if __name__ == "__main__": main() | 90 |
91 if __name__ == "__main__": | |
92 main() |