annotate data_manager/data_manager_snpEff_databases.py @ 0:e1d9f6a0ad53

Uploaded
author jjohnson
date Thu, 04 Jul 2013 10:43:55 -0400
parents
children 6ad9205c1307
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
1 #!/usr/bin/env python
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
2
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
3 import sys
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
4 import os
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
5 import re
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
6 import tempfile
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
7 import subprocess
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
8 import fileinput
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
9 import shutil
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
10 import optparse
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
11 import urllib2
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
12 from ftplib import FTP
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
13 import tarfile
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
14
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
15 from galaxy.util.json import from_json_string, to_json_string
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
16
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
17 def stop_err(msg):
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
18 sys.stderr.write(msg)
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
19 sys.exit(1)
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
20
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
21 def fetch_databases(data_manager_dict, target_directory, jar_path):
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
22 (snpEff_dir,snpEff_jar) = os.path.split(jar_path)
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
23 if not os.path.exists(target_directory):
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
24 os.makedirs(target_directory)
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
25 databases_path = os.path.join( target_directory, 'databases.out' )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
26 databases_output = open(databases_path,'w')
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
27 args = [ 'java','-jar', ]
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
28 args.append( snpEff_jar )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
29 args.append( 'databases' )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
30 # tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-snpEff-stderr" )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
31 # databases_output = open(databases_path)
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
32 # proc = subprocess.Popen( args=args, shell=False, cwd=snpEff_dir, stdout=databases_output.fileno(), stderr=tmp_stderr.fileno() )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
33 proc = subprocess.Popen( args=args, shell=False, cwd=snpEff_dir, stdout=databases_output.fileno() )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
34 return_code = proc.wait()
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
35 if return_code:
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
36 sys.exit( return_code )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
37 databases_output.close()
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
38 try:
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
39 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
40 data_manager_dict['data_tables']['snpeff_databases'] = data_manager_dict['data_tables'].get( 'snpeff_databases', [] )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
41 data_table_entries = []
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
42 fh = open(databases_path,'r')
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
43 for i,line in enumerate(fh):
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
44 fields = line.split('\t')
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
45 if len(fields) >= 2:
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
46 genome_version = fields[0].strip()
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
47 if genome_version.startswith("Genome") or genome_version.startswith("-"):
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
48 continue
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
49 description = fields[1].strip()
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
50 data_table_entries.append(dict(value=genome_version, name=description))
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
51 data_manager_dict['data_tables']['snpeff_databases'] = data_table_entries
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
52 except Exception, e:
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
53 stop_err( 'Error parsing %s %s\n' % (config,str( e )) )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
54 else:
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
55 fh.close()
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
56 return data_manager_dict
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
57
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
58 def main():
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
59 #Parse Command Line
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
60 parser = optparse.OptionParser()
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
61 parser.add_option( '-j', '--jar_path', dest='jar_path', action='store', type="string", default=None, help='snpEff.jar path' )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
62 (options, args) = parser.parse_args()
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
63
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
64 filename = args[0]
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
65
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
66 params = from_json_string( open( filename ).read() )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
67 target_directory = params[ 'output_data' ][0]['extra_files_path']
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
68 os.mkdir( target_directory )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
69 data_manager_dict = {}
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
70
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
71
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
72 #Create Defuse Reference Data
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
73 data_manager_dict = fetch_databases( data_manager_dict, target_directory, options.jar_path)
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
74
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
75 #save info to json file
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
76 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) )
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
77
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
78 if __name__ == "__main__": main()
e1d9f6a0ad53 Uploaded
jjohnson
parents:
diff changeset
79