19
|
1 #!/usr/bin/env python
|
|
2
|
|
3 import sys, os, optparse, shutil
|
|
4
|
|
5 __version__ = "1.0.0"
|
|
6 VALID_CHARS = '.-()[]0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
|
|
7 # in the case of collections, exitcodes equal to 0 and 1 are not considered errors
|
|
8 ERR_EXIT_CODE = 2
|
|
9 OK_EXIT_CODE = 0
|
|
10
|
|
11 def printLog( logfilepath, message, exitcode=OK_EXIT_CODE, exit=False ):
|
|
12 print message
|
|
13 with open( logfilepath, 'a+' ) as out:
|
|
14 out.write( message + '\n' )
|
|
15 if exit:
|
|
16 sys.exit( exitcode )
|
|
17
|
|
18 def querySBT( options, args ):
|
|
19 output_dir_path = options.outputdir
|
|
20 outlogfile = options.outfile
|
|
21
|
|
22 tree_file_paths = options.treep.split( ',' )
|
|
23 tree_file_names = options.treen.split( ',' )
|
|
24 tree_def_filepath = None
|
|
25 leafnames_filepath = None
|
|
26 for idx, tree_file_name in enumerate( tree_file_names ):
|
|
27 if tree_file_name == 'howde':
|
|
28 tree_def_filepath = tree_file_paths[ idx ]
|
|
29 elif tree_file_name == 'leafnames':
|
|
30 leafnames_filepath = tree_file_paths[ idx ]
|
|
31 if tree_def_filepath is not None and leafnames_filepath is not None:
|
|
32 break
|
|
33
|
|
34 if tree_def_filepath is not None and leafnames_filepath is not None:
|
|
35 leafnames_counter = 0
|
|
36 with open( leafnames_filepath ) as leafnames_file:
|
|
37 for line in leafnames_file:
|
|
38 if line.strip():
|
|
39 leafnames_counter += 1
|
|
40 if leafnames_counter > 0:
|
|
41 printLog( outlogfile, 'The selected collection contains a valid tree' )
|
|
42 shutil.copyfile( tree_def_filepath, 'howde.txt' )
|
|
43 tree_def_filepath = 'howde.txt'
|
|
44 for idx, tree_file_name in enumerate( tree_file_names ):
|
|
45 if tree_file_name.endswith( 'detbrief.rrr' ):
|
|
46 shutil.copyfile( tree_file_paths[ idx ], tree_file_name + '.bf' )
|
|
47
|
|
48 printLog( outlogfile, 'Creating batch of queries' )
|
|
49 # create tmp batch file
|
|
50 batch_file_name = 'queries.fa'
|
|
51 batch_file = open( batch_file_name, 'w' )
|
|
52
|
|
53 comma_sep_file_paths = options.files
|
|
54 # check if options.files contains at least one file path
|
|
55 if comma_sep_file_paths is not None:
|
|
56 # split file paths
|
|
57 file_paths = comma_sep_file_paths.split(",")
|
|
58 # split file names
|
|
59 file_names = options.names.split(",")
|
|
60 for idx, file_path in enumerate(file_paths):
|
|
61 fixed_file_name = ''.join( c for c in file_names[ idx ] if c in VALID_CHARS )
|
|
62 printLog( outlogfile, '> processing file ' + file_names[ idx ] + ' ( fixed_name=\"' + fixed_file_name + '\" ) ' )
|
|
63 with open(file_path, 'r') as content_file:
|
|
64 for line in content_file:
|
|
65 line = line.strip()
|
|
66 if line:
|
|
67 line_split = line.strip().split("\t") # split on tab
|
|
68 if len(line_split) == 2: # 0:id , 1:seq , otherwise skip line
|
|
69 original_seq_id = line_split[0]
|
|
70 # fix seq_id using valid chars only
|
|
71 seq_id = ''.join( c for c in original_seq_id if c in VALID_CHARS )
|
|
72 printLog( outlogfile, '> sequence ' + original_seq_id + ' ( fixed_name=\"' + seq_id + '\" )' )
|
|
73 seq_text = line_split[1]
|
|
74
|
|
75 # write on batch file
|
|
76 batch_file.write( '> ' + fixed_file_name + '_' + seq_id + '\n' + seq_text + '\n' )
|
|
77 batch_file.close()
|
|
78 # query the tree
|
|
79 printLog( outlogfile, 'Querying the tree' )
|
|
80 query_res_file_path = os.path.abspath( 'answer.txt' )
|
|
81 sort_param = '--sort'
|
|
82 if options.sort == 0:
|
|
83 sort_param = ''
|
21
|
84 query_exitcode = os.system( 'howdesbt query --tree=' + os.path.abspath( tree_def_filepath ) + ' ' + os.path.abspath( batch_file_name ) + '=' + str(options.threshold) + ' --out=' + query_res_file_path + ' ' + sort_param )
|
19
|
85 if query_exitcode > 0:
|
|
86 printLog( outlogfile, '> ERROR: an error has occurred while querying the tree with the sequence [id: ' + seq_id + '] in input file ' + file_names[ idx ] )
|
|
87 else:
|
|
88 if os.path.exists( query_res_file_path ):
|
|
89 with open( query_res_file_path ) as query_res_file:
|
|
90 file_path = ''
|
|
91 theta_matches = 0
|
|
92 for line in query_res_file:
|
|
93 line = line.strip()
|
|
94 if line:
|
|
95 if line.startswith( '*' ):
|
|
96 line_split = line.split( ' ' )
|
|
97 theta_matches = int( line_split[ 1 ] )
|
|
98 file_name = line_split[ 0 ].replace( '*', '' )
|
|
99 file_path = os.path.join( output_dir_path, file_name + '_txt' )
|
|
100 open( file_path, 'a' ).close()
|
|
101 else:
|
|
102 res_file = open( file_path, 'a+' )
|
|
103 fraction = str( theta_matches ) + '/' + str( leafnames_counter )
|
|
104 score = format( round( float( theta_matches ) / float( leafnames_counter ) , 6 ), '6f' )
|
|
105 res_file.write( line + '\t' + fraction + '\t' + score + '\n' )
|
|
106 res_file.close()
|
|
107 else:
|
|
108 printLog( outlogfile, 'An error has occurred while querying the tree', exitcode=ERR_EXIT_CODE, exit=True )
|
|
109 else:
|
|
110 printLog( outlogfile, 'The selected collection does not contain a valid tree', exitcode=ERR_EXIT_CODE, exit=True )
|
|
111 else:
|
|
112 printLog( outlogfile, 'The selected collection does not contain a valid tree', exitcode=ERR_EXIT_CODE, exit=True )
|
|
113
|
|
114 def __main__():
|
|
115 # Parse the command line options
|
|
116 usage = "Usage: query.py --files comma_sep_file_paths --names comma_seq_file_names --sequences sequences_text --search search_mode --exact exact_alg --sthreshold threshold --outputdir output_dir_path"
|
|
117 parser = optparse.OptionParser(usage = usage)
|
|
118 parser.add_option("-v", "--version", action="store_true", dest="version",
|
|
119 default=False, help="display version and exit")
|
|
120 parser.add_option("-f", "--files", type="string",
|
|
121 action="store", dest="files", help="comma separated files path")
|
|
122 parser.add_option("-n", "--names", type="string",
|
|
123 action="store", dest="names", help="comma separated names associated to the files specified in --files")
|
|
124 parser.add_option("-k", "--treep", type="string",
|
|
125 action="store", dest="treep", help="paths of files in collection")
|
|
126 parser.add_option("-m", "--treen", type="string",
|
|
127 action="store", dest="treen", help="names of files in collection")
|
|
128 parser.add_option("-t", "--threshold", type="float", default=0.7,
|
|
129 action="store", dest="threshold", help="search threshold")
|
|
130 parser.add_option("-s", "--sort", type="int", default=1,
|
|
131 action="store", dest="sort", help="sort results")
|
|
132 parser.add_option("-o", "--outputdir", type="string", default="output",
|
|
133 action="store", dest="outputdir", help="output directory (collection) path")
|
|
134 parser.add_option("-r", "--outfile", type="string", default="query.txt",
|
|
135 action="store", dest="outfile", help="output log file path")
|
|
136
|
|
137 (options, args) = parser.parse_args()
|
|
138 if options.version:
|
|
139 print __version__
|
|
140 else:
|
|
141 # create output dir (collection)
|
|
142 output_dir_path = options.outputdir
|
|
143 if not os.path.exists(output_dir_path):
|
|
144 os.makedirs(output_dir_path)
|
|
145
|
|
146 querySBT( options, args )
|
|
147
|
|
148 if __name__ == "__main__": __main__()
|