annotate data_manager/bowtie2_index_builder.py @ 3:49331eedb1c5 draft

planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_bowtie2_index_builder commit c54b039220cc3d1767a424b6ae76852d40e67437
author devteam
date Mon, 16 Nov 2015 15:12:06 -0500
parents 8de13c28d92e
children 8a892275932f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
1 #!/usr/bin/env python
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
2 #Dan Blankenberg
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
3
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
4 import sys
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
5 import os
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
6 import optparse
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
7 import subprocess
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
8
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
9 from json import loads, dumps
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
10
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
11
2
8de13c28d92e planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents: 1
diff changeset
12 DEFAULT_DATA_TABLE_NAMES = ["bowtie2_indexes"]
0
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
13
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
14 def get_id_name( params, dbkey, fasta_description=None):
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
15 #TODO: ensure sequence_id is unique and does not already appear in location file
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
16 sequence_id = params['param_dict']['sequence_id']
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
17 if not sequence_id:
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
18 sequence_id = dbkey
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
19
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
20 sequence_name = params['param_dict']['sequence_name']
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
21 if not sequence_name:
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
22 sequence_name = fasta_description
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
23 if not sequence_name:
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
24 sequence_name = dbkey
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
25 return sequence_id, sequence_name
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
26
2
8de13c28d92e planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents: 1
diff changeset
27 def build_bowtie2_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_names=DEFAULT_DATA_TABLE_NAMES ):
0
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
28 #TODO: allow multiple FASTA input files
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
29 fasta_base_name = os.path.split( fasta_filename )[-1]
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
30 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
31 os.symlink( fasta_filename, sym_linked_fasta_filename )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
32 args = [ 'bowtie2-build', sym_linked_fasta_filename, sequence_id ]
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
33 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
34 return_code = proc.wait()
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
35 if return_code:
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
36 print >> sys.stderr, "Error building index."
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
37 sys.exit( return_code )
1
e0839cffd009 Uploaded
devteam
parents: 0
diff changeset
38 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=sequence_id )
2
8de13c28d92e planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents: 1
diff changeset
39 for data_table_name in data_table_names:
8de13c28d92e planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents: 1
diff changeset
40 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
0
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
41
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
42 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
43 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
44 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
45 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
46 return data_manager_dict
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
47
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
48 def main():
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
49 #Parse Command Line
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
50 parser = optparse.OptionParser()
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
51 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
52 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
53 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
2
8de13c28d92e planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents: 1
diff changeset
54 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='append', type="string", default=None, help='data_table_name' )
0
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
55 (options, args) = parser.parse_args()
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
56
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
57 filename = args[0]
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
58
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
59 params = loads( open( filename ).read() )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
60 target_directory = params[ 'output_data' ][0]['extra_files_path']
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
61 os.mkdir( target_directory )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
62 data_manager_dict = {}
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
63
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
64 dbkey = options.fasta_dbkey
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
65
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
66 if dbkey in [ None, '', '?' ]:
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
67 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
68
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
69 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
70
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
71 #build the index
2
8de13c28d92e planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents: 1
diff changeset
72 build_bowtie2_index( data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_names=options.data_table_name or DEFAULT_DATA_TABLE_NAMES )
0
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
73
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
74 #save info to json file
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
75 open( filename, 'wb' ).write( dumps( data_manager_dict ) )
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
76
eec64de5f7fb Uploaded
devteam
parents:
diff changeset
77 if __name__ == "__main__": main()