annotate data_manager/bowtie_index_builder.py @ 0:119b03a6eaaa draft default tip

Uploaded
author jjohnson
date Tue, 20 Aug 2013 11:55:20 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
1 #!/usr/bin/env python
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
2
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
3 import sys
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
4 import os
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
5 import tempfile
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
6 import optparse
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
7 import subprocess
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
8
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
9 from galaxy.util.json import from_json_string, to_json_string
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
10
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
11
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
12 CHUNK_SIZE = 2**20
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
13
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
14 DEFAULT_DATA_TABLE_NAME = "bowtie_indexes"
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
15
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
16 def get_id_name( params, dbkey, fasta_description=None):
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
17 #TODO: ensure sequence_id is unique and does not already appear in location file
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
18 sequence_id = params['param_dict']['sequence_id']
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
19 if not sequence_id:
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
20 sequence_id = dbkey
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
21
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
22 sequence_name = params['param_dict']['sequence_name']
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
23 if not sequence_name:
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
24 sequence_name = fasta_description
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
25 if not sequence_name:
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
26 sequence_name = dbkey
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
27 return sequence_id, sequence_name
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
28
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
29 def build_bowtie_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME, color_space = False ):
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
30 #TODO: allow multiple FASTA input files
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
31 #tmp_dir = tempfile.mkdtemp( prefix='tmp-data-manager-bowtie-index-builder-' )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
32 fasta_base_name = os.path.split( fasta_filename )[-1]
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
33 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
34 os.symlink( fasta_filename, sym_linked_fasta_filename )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
35 args = [ 'bowtie-build' ]
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
36 if color_space:
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
37 args.append( '-c' )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
38 args.append( sym_linked_fasta_filename)
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
39 args.append( fasta_base_name )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
40 args.append( sym_linked_fasta_filename )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
41 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-bowtie-index-builder-stderr" )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
42 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
43 return_code = proc.wait()
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
44 if return_code:
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
45 tmp_stderr.flush()
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
46 tmp_stderr.seek(0)
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
47 print >> sys.stderr, "Error building index:"
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
48 while True:
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
49 chunk = tmp_stderr.read( CHUNK_SIZE )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
50 if not chunk:
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
51 break
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
52 sys.stderr.write( chunk )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
53 sys.exit( return_code )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
54 tmp_stderr.close()
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
55 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
56 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
57
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
58 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
59 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
60 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
61 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
62 return data_manager_dict
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
63
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
64 def main():
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
65 #Parse Command Line
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
66 parser = optparse.OptionParser()
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
67 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
68 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
69 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
70 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
71 parser.add_option( '-c', '--color_space', dest='color_space', action='store_true', default=False, help='color_space' )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
72 (options, args) = parser.parse_args()
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
73
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
74 filename = args[0]
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
75
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
76 params = from_json_string( open( filename ).read() )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
77 target_directory = params[ 'output_data' ][0]['extra_files_path']
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
78 os.mkdir( target_directory )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
79 data_manager_dict = {}
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
80
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
81 dbkey = options.fasta_dbkey
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
82
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
83 if dbkey in [ None, '', '?' ]:
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
84 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
85
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
86 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
87
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
88 #build the index
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
89 build_bowtie_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 )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
90
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
91 #save info to json file
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
92 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) )
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
93
119b03a6eaaa Uploaded
jjohnson
parents:
diff changeset
94 if __name__ == "__main__": main()