annotate data_manager/data_manager_sam_fasta_index_builder.py @ 0:d6b6db69d9bf draft

Uploaded data manager definition.
author devteam
date Thu, 27 Mar 2014 14:15:02 -0400
parents
children 2a1ac1abc3f7
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
1 #!/usr/bin/env python
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
2 #Dan Blankenberg
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
3
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
4 import json
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
5 import optparse
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
6 import os
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
7 import subprocess
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
8 import sys
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
9 import tempfile
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
10
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
11 CHUNK_SIZE = 2**20
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
12
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
13 DEFAULT_DATA_TABLE_NAME = "fasta_indexes"
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
14
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
15 def get_id_name( params, dbkey, fasta_description=None):
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
16 #TODO: ensure sequence_id is unique and does not already appear in location file
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
17 sequence_id = params['param_dict']['sequence_id']
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
18 if not sequence_id:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
19 sequence_id = dbkey
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
20
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
21 sequence_name = params['param_dict']['sequence_name']
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
22 if not sequence_name:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
23 sequence_name = fasta_description
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
24 if not sequence_name:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
25 sequence_name = dbkey
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
26 return sequence_id, sequence_name
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
27
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
28 def build_sam_index( data_manager_dict, fasta_filename, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME ):
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
29 #TODO: allow multiple FASTA input files
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
30 fasta_base_name = os.path.split( fasta_filename )[-1]
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
31 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
32 os.symlink( fasta_filename, sym_linked_fasta_filename )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
33
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
34 args = [ 'samtools', 'faidx' ]
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
35 args.append( sym_linked_fasta_filename )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
36 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-sam_fa_index_builder-stderr" )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
37 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
38 return_code = proc.wait()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
39 if return_code:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
40 tmp_stderr.flush()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
41 tmp_stderr.seek( 0 )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
42 sys.stderr.write( "Error building index:\n" )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
43 while True:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
44 chunk = tmp_stderr.read( CHUNK_SIZE )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
45 if not chunk:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
46 break
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
47 sys.stderr.write( chunk )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
48 sys.exit( return_code )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
49 tmp_stderr.close()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
50 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
51 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
52
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
53 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
54 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
55 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
56 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
57 return data_manager_dict
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
58
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
59 def main():
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
60 #Parse Command Line
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
61 parser = optparse.OptionParser()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
62 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
63 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
64 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
65 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
66 (options, args) = parser.parse_args()
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
67
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
68 filename = args[0]
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
69
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
70 params = json.loads( open( filename ).read() )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
71 target_directory = params[ 'output_data' ][0]['extra_files_path']
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
72 os.mkdir( target_directory )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
73 data_manager_dict = {}
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
74
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
75 if options.fasta_dbkey in [ None, '', '?' ]:
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
76 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( options.fasta_dbkey ) )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
77
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
78 sequence_id, sequence_name = get_id_name( params, dbkey=options.fasta_dbkey, fasta_description=options.fasta_description )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
79
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
80 #build the index
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
81 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 )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
82
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
83 #save info to json file
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
84 open( filename, 'wb' ).write( json.dumps( data_manager_dict ) )
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
85
d6b6db69d9bf Uploaded data manager definition.
devteam
parents:
diff changeset
86 if __name__ == "__main__": main()