|
0
|
1 #!/usr/bin/env python
|
|
|
2 # Dan Blankenberg
|
|
|
3 from __future__ import print_function
|
|
|
4
|
|
|
5 import optparse
|
|
|
6 import os
|
|
|
7 import subprocess
|
|
|
8 import sys
|
|
|
9 from json import dumps, loads
|
|
|
10
|
|
|
11 CHUNK_SIZE = 2**20
|
|
|
12 TWO_GB = 2**30 * 2
|
|
|
13 DEFAULT_DATA_TABLE_NAME = "mummer4_indexes"
|
|
|
14
|
|
|
15
|
|
|
16 def get_id_name( params, dbkey, fasta_description=None):
|
|
|
17 # TODO: ensure sequence_id is unique and does not already appear in location file
|
|
|
18 sequence_id = params['param_dict']['sequence_id']
|
|
|
19 if not sequence_id:
|
|
|
20 sequence_id = dbkey
|
|
|
21
|
|
|
22 sequence_name = params['param_dict']['sequence_name']
|
|
|
23 if not sequence_name:
|
|
|
24 sequence_name = fasta_description
|
|
|
25 if not sequence_name:
|
|
|
26 sequence_name = dbkey
|
|
|
27 return sequence_id, sequence_name
|
|
|
28
|
|
|
29
|
|
|
30 def build_mummer4_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME ):
|
|
|
31 # TODO: allow multiple FASTA input files
|
|
|
32 fasta_base_name = os.path.split( fasta_filename )[-1]
|
|
|
33 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
|
|
|
34 os.symlink( fasta_filename, sym_linked_fasta_filename )
|
|
|
35 if params['param_dict']['index_algorithm'] == 'automatic':
|
|
|
36 if os.stat( fasta_filename ).st_size < TWO_GB: # use 2 GB as cut off for memory vs. max of 2gb database size; this is somewhat arbitrary
|
|
|
37 index_algorithm = 'is'
|
|
|
38 else:
|
|
|
39 index_algorithm = 'bwtsw'
|
|
|
40 else:
|
|
|
41 index_algorithm = params['param_dict']['index_algorithm']
|
|
|
42
|
|
|
43 args = [ 'mummer4', '-save', dbkey, sym_linked_fasta_filename, sym_linked_fasta_filename ]
|
|
|
44 args.append( sym_linked_fasta_filename )
|
|
|
45 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory )
|
|
|
46 return_code = proc.wait()
|
|
|
47 if return_code:
|
|
|
48 print("Error building index.", file=sys.stderr)
|
|
|
49 sys.exit( return_code )
|
|
|
50 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
|
|
|
51 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
|
|
|
52
|
|
|
53
|
|
|
54 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', {} )
|
|
|
56 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 )
|
|
|
58 return data_manager_dict
|
|
|
59
|
|
|
60
|
|
|
61 def main():
|
|
|
62 parser = optparse.OptionParser()
|
|
|
63 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' )
|
|
|
65 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' )
|
|
|
67 (options, args) = parser.parse_args()
|
|
|
68
|
|
|
69 filename = args[0]
|
|
|
70
|
|
|
71 params = loads( open( filename ).read() )
|
|
|
72 target_directory = params[ 'output_data' ][0]['extra_files_path']
|
|
|
73 os.mkdir( target_directory )
|
|
|
74 data_manager_dict = {}
|
|
|
75
|
|
|
76 dbkey = options.fasta_dbkey
|
|
|
77
|
|
|
78 if dbkey in [ None, '', '?' ]:
|
|
|
79 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
|
|
|
80
|
|
|
81 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
|
|
|
82
|
|
|
83 # build the index
|
|
|
84 build_mummer4_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 )
|
|
|
85
|
|
|
86 # save info to json file
|
|
|
87 open( filename, 'wb' ).write( dumps( data_manager_dict ) )
|
|
|
88
|
|
|
89
|
|
|
90 if __name__ == "__main__":
|
|
|
91 main()
|
