|
0
|
1 #!/usr/bin/env python
|
|
|
2
|
|
|
3 import sys
|
|
|
4 import os
|
|
|
5 import re
|
|
|
6 import tempfile
|
|
|
7 import subprocess
|
|
|
8 import fileinput
|
|
|
9 import shutil
|
|
|
10 import optparse
|
|
|
11 import urllib2
|
|
|
12 from ftplib import FTP
|
|
|
13 import tarfile
|
|
|
14
|
|
|
15 from galaxy.util.json import from_json_string, to_json_string
|
|
|
16
|
|
|
17 def stop_err(msg):
|
|
|
18 sys.stderr.write(msg)
|
|
|
19 sys.exit(1)
|
|
|
20
|
|
|
21 """
|
|
|
22 # Download human database 'hg19'
|
|
|
23 java -jar snpEff.jar download -v hg19
|
|
|
24
|
|
2
|
25 <command>java -jar \$SNPEFF_JAR_PATH/snpEff.jar download -c \$JAVA_JAR_PATH/snpEff.config $genomeVersion > $logfile </command>
|
|
0
|
26
|
|
|
27 snpEffectPredictor.bin
|
|
|
28 regulation_HeLa-S3.bin
|
|
|
29 regulation_pattern = 'regulation_(.+).bin'
|
|
|
30
|
|
|
31
|
|
|
32 """
|
|
|
33 def download_database(data_manager_dict, target_directory, jar_path,config,genome_version,organism):
|
|
|
34 ## get data_dir from config
|
|
|
35 ##---
|
|
|
36 ## Databases are stored here
|
|
|
37 ## E.g.: Information for 'hg19' is stored in data_dir/hg19/
|
|
|
38 ##
|
|
|
39 ## Note: Since version 2.1 you can use tilde ('~') as first character to refer to your home directory
|
|
|
40 ##---
|
|
|
41 #data_dir = ~/snpEff/data/
|
|
|
42 data_dir = None
|
|
|
43 try:
|
|
|
44 fh = open(config)
|
|
|
45 for i,line in enumerate(fh):
|
|
|
46 if line.strip().startswith('data_dir'):
|
|
|
47 (k,v) = line.split('=')
|
|
|
48 data_dir = os.path.expanduser(v.strip())
|
|
|
49 break
|
|
|
50 except Exception, e:
|
|
|
51 stop_err( 'Error parsing %s %s\n' % (config,str( e )) )
|
|
|
52 else:
|
|
|
53 fh.close()
|
|
|
54 (snpEff_dir,snpEff_jar) = os.path.split(jar_path)
|
|
|
55 args = [ 'java','-jar' ]
|
|
|
56 args.append( jar_path )
|
|
|
57 args.append( 'download' )
|
|
|
58 # args.append( '-c' )
|
|
|
59 # args.append( 'config' )
|
|
|
60 args.append( '-v' )
|
|
|
61 args.append( genome_version )
|
|
|
62 proc = subprocess.Popen( args=args, shell=False, cwd=snpEff_dir )
|
|
|
63 return_code = proc.wait()
|
|
|
64 if return_code:
|
|
|
65 sys.exit( return_code )
|
|
|
66 ## search data_dir/genome_version for files
|
|
|
67 regulation_pattern = 'regulation_(.+).bin'
|
|
2
|
68 # annotation files that are included in snpEff by a flag
|
|
|
69 annotations_dict = {'nextProt.bin' : '-nextprot','motif.bin': '-motif'}
|
|
0
|
70 genome_path = os.path.join(data_dir,genome_version)
|
|
|
71 if os.path.isdir(genome_path):
|
|
|
72 for root, dirs, files in os.walk(genome_path):
|
|
|
73 for fname in files:
|
|
|
74 if fname.startswith('snpEffectPredictor'):
|
|
|
75 # if snpEffectPredictor.bin download succeeded
|
|
|
76 name = genome_version + (' : ' + organism if organism else '')
|
|
|
77 data_table_entry = dict(value=genome_version, name=name)
|
|
|
78 _add_data_table_entry( data_manager_dict, 'snpeff_genomedb', data_table_entry )
|
|
|
79 else:
|
|
|
80 m = re.match(regulation_pattern,fname)
|
|
|
81 if m:
|
|
|
82 name = m.groups()[0]
|
|
2
|
83 data_table_entry = dict(genome=genome_version,value=name, name=name)
|
|
0
|
84 _add_data_table_entry( data_manager_dict, 'snpeff_regulationdb', data_table_entry )
|
|
2
|
85 elif fname in annotations_dict:
|
|
|
86 value = annotations_dict[fname]
|
|
|
87 name = value.lstrip('-')
|
|
|
88 data_table_entry = dict(genome=genome_version,value=value, name=name)
|
|
|
89 _add_data_table_entry( data_manager_dict, 'snpeff_annotations', data_table_entry )
|
|
0
|
90 return data_manager_dict
|
|
|
91
|
|
|
92 def _add_data_table_entry( data_manager_dict, data_table, data_table_entry ):
|
|
|
93 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
|
|
|
94 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get( data_table, [] )
|
|
|
95 data_manager_dict['data_tables'][data_table].append( data_table_entry )
|
|
|
96 return data_manager_dict
|
|
|
97
|
|
|
98 def main():
|
|
|
99 #Parse Command Line
|
|
|
100 parser = optparse.OptionParser()
|
|
|
101 parser.add_option( '-j', '--jar_path', dest='jar_path', action='store', type="string", default=None, help='snpEff.jar path' )
|
|
|
102 parser.add_option( '-c', '--config', dest='config', action='store', type="string", default=None, help='snpEff.config path' )
|
|
|
103 parser.add_option( '-g', '--genome_version', dest='genome_version', action='store', type="string", default=None, help='genome_version' )
|
|
|
104 parser.add_option( '-o', '--organism', dest='organism', action='store', type="string", default=None, help='organism name' )
|
|
|
105 (options, args) = parser.parse_args()
|
|
|
106
|
|
|
107 filename = args[0]
|
|
|
108
|
|
|
109 params = from_json_string( open( filename ).read() )
|
|
|
110 target_directory = params[ 'output_data' ][0]['extra_files_path']
|
|
|
111 os.mkdir( target_directory )
|
|
|
112 data_manager_dict = {}
|
|
|
113
|
|
|
114
|
|
|
115 #Create SnpEff Reference Data
|
|
|
116 download_database(data_manager_dict, target_directory, options.jar_path,options.config,options.genome_version,options.organism)
|
|
|
117
|
|
|
118 #save info to json file
|
|
|
119 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) )
|
|
|
120
|
|
|
121 if __name__ == "__main__": main()
|
|
|
122
|
