annotate data_manager/picard_index_builder.py @ 0:ce8e6a5b8830 draft default tip

Uploaded
author jjohnson
date Tue, 20 Aug 2013 11:12:02 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
1 #!/usr/bin/env python
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
2 #Jim Johnson
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
3
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
4 import sys
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
5 import os
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
6 import tempfile
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
7 import optparse
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
8 import subprocess
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
9
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
10 from galaxy.util.json import from_json_string, to_json_string
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
11
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
12 CHUNK_SIZE = 2**20
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
13
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
14 DEFAULT_DATA_TABLE_NAME = "picard_indexes"
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
15
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
16 def get_id_name( params, dbkey, fasta_description=None):
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
17 #TODO: ensure sequence_id is unique and does not already appear in location file
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
18 sequence_id = params['param_dict']['sequence_id']
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
19 if not sequence_id:
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
20 sequence_id = dbkey
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
21
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
22 sequence_name = params['param_dict']['sequence_name']
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
23 if not sequence_name:
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
24 sequence_name = fasta_description
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
25 if not sequence_name:
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
26 sequence_name = dbkey
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
27 return sequence_id, sequence_name
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
28
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
29 def build_picard_index( jar_path, data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME):
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
30 #TODO: allow multiple FASTA input files
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
31 #tmp_dir = tempfile.mkdtemp( prefix='tmp-data-manager-picard-index-builder-' )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
32 fasta_base_name = os.path.split( fasta_filename )[-1]
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
33 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
34 os.symlink( fasta_filename, sym_linked_fasta_filename )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
35 basename,ext = os.path.splitext(fasta_base_name)
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
36 ref_dict = "%s.dict" % basename
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
37 args = [ 'java','-Xmx4G','-jar', jar_path, "R=%s" % sym_linked_fasta_filename, "O=%s" % ref_dict ]
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
38 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-picard-index-builder-stderr" )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
39 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
40 return_code = proc.wait()
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
41 if return_code:
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
42 tmp_stderr.flush()
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
43 tmp_stderr.seek(0)
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
44 print >> sys.stderr, "Error building index:"
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
45 while True:
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
46 chunk = tmp_stderr.read( CHUNK_SIZE )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
47 if not chunk:
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
48 break
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
49 sys.stderr.write( chunk )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
50 sys.exit( return_code )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
51 tmp_stderr.close()
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
52 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
53 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
54
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
55 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
56 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
57 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
58 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
59 return data_manager_dict
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
60
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
61 def main():
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
62 #Parse Command Line
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
63 parser = optparse.OptionParser()
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
64 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
65 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
66 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
67 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
68 parser.add_option( '-j', '--jar_path', dest='jar_path', action='store', type="string", default=None, help='path to CreateSequenceDictionary.jar' )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
69 (options, args) = parser.parse_args()
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
70
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
71 filename = args[0]
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
72
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
73 params = from_json_string( open( filename ).read() )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
74 target_directory = params[ 'output_data' ][0]['extra_files_path']
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
75 os.mkdir( target_directory )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
76 data_manager_dict = {}
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
77
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
78 dbkey = options.fasta_dbkey
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
79
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
80 if dbkey in [ None, '', '?' ]:
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
81 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
82
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
83 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
84
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
85 #build the index
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
86 build_picard_index( options.jar_path, 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 )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
87
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
88 #save info to json file
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
89 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) )
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
90
ce8e6a5b8830 Uploaded
jjohnson
parents:
diff changeset
91 if __name__ == "__main__": main()