Mercurial > repos > simonl > agile_wrapper
comparison Users/oconnorlab/Desktop/agile/agile_wrapper.py @ 0:d6a426afaa46 default tip
Migrated tool version 1.0.0 from old tool shed archive to new tool shed repository
author | simonl |
---|---|
date | Tue, 07 Jun 2011 16:22:51 -0400 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:d6a426afaa46 |
---|---|
1 #!/usr/bin/env python | |
2 | |
3 import os, sys, tempfile | |
4 | |
5 assert sys.version_info[:2] >= (2.4) | |
6 | |
7 def stop_err( msg ): | |
8 sys.stderr.write( "%s\n" % msg ) | |
9 sys.exit() | |
10 | |
11 def check_nib_file( dbkey, GALAXY_DATA_INDEX_DIR ): | |
12 nib_file = "%s/alignseq.loc" % GALAXY_DATA_INDEX_DIR | |
13 nib_path = '' | |
14 nibs = {} | |
15 for i, line in enumerate( file( nib_file ) ): | |
16 line = line.rstrip( '\r\n' ) | |
17 if line and not line.startswith( "#" ): | |
18 fields = line.split( '\t' ) | |
19 if len( fields ) < 3: | |
20 continue | |
21 if fields[0] == 'seq': | |
22 nibs[( fields[1] )] = fields[2] | |
23 if nibs.has_key( dbkey ): | |
24 nib_path = nibs[( dbkey )] | |
25 return nib_path | |
26 | |
27 def check_twobit_file( dbkey, GALAXY_DATA_INDEX_DIR ): | |
28 twobit_file = "%s/twobit.loc" % GALAXY_DATA_INDEX_DIR | |
29 twobit_path = '' | |
30 twobits = {} | |
31 for i, line in enumerate( file( twobit_file ) ): | |
32 line = line.rstrip( '\r\n' ) | |
33 if line and not line.startswith( "#" ): | |
34 fields = line.split( '\t' ) | |
35 if len( fields ) < 2: | |
36 continue | |
37 twobits[( fields[0] )] = fields[1] | |
38 if twobits.has_key( dbkey ): | |
39 twobit_path = twobits[( dbkey )] | |
40 return twobit_path | |
41 | |
42 def __main__(): | |
43 # I/O | |
44 source_format = sys.argv[1] # 0: dbkey; 1: upload file | |
45 target_file = sys.argv[2] | |
46 query_file = sys.argv[3] | |
47 output_file = sys.argv[4] | |
48 max_sims = sys.argv[5] | |
49 tile_size = sys.argv[6] | |
50 max_freq = sys.argv[7] | |
51 out_type = sys.argv[8] | |
52 all_match = sys.argv[9] | |
53 | |
54 GALAXY_DATA_INDEX_DIR = sys.argv[10] | |
55 | |
56 all_files = [] | |
57 if source_format == '0': | |
58 # check target genome | |
59 dbkey = target_file | |
60 nib_path = check_nib_file( dbkey, GALAXY_DATA_INDEX_DIR ) | |
61 twobit_path = check_twobit_file( dbkey, GALAXY_DATA_INDEX_DIR ) | |
62 if not os.path.exists( nib_path ) and not os.path.exists( twobit_path ): | |
63 stop_err("No sequences are available for %s, request them by reporting this error." % dbkey) | |
64 | |
65 # check the query file, see whether all of them are legitimate sequence | |
66 if nib_path and os.path.isdir( nib_path ): | |
67 compress_files = os.listdir(nib_path) | |
68 target_path = nib_path | |
69 elif twobit_path: | |
70 compress_files = [twobit_path] | |
71 target_path = "" | |
72 else: | |
73 stop_err("Requested genome build has no available sequence.") | |
74 | |
75 for file in compress_files: | |
76 file = "%s/%s" % ( target_path, file ) | |
77 file = os.path.normpath(file) | |
78 all_files.append(file) | |
79 else: | |
80 all_files = [target_file] | |
81 | |
82 for detail_file_path in all_files: | |
83 output_tempfile = tempfile.NamedTemporaryFile().name | |
84 if all_match == "true": | |
85 command = "agile %s %s -maxSIMs=%s -tileSize=%s -maxFreq=%s -out=%s -all %s 2>&1" % ( detail_file_path, query_file, max_sims, tile_size, max_freq, out_type, output_tempfile ) | |
86 else: | |
87 command = "agile %s %s -maxSIMs=%s -tileSize=%s -maxFreq=%s -out=%s %s 2>&1" % ( detail_file_path, query_file, max_sims, tile_size, max_freq, out_type, output_tempfile ) | |
88 | |
89 os.system( command ) | |
90 os.system( 'cat %s >> %s' % ( output_tempfile, output_file ) ) | |
91 os.remove( output_tempfile ) | |
92 | |
93 if __name__ == '__main__': __main__() |