annotate select_restriction_enzymes.py @ 0:939c20789501 default tip

Uploaded
author rico
date Fri, 06 Apr 2012 10:51:28 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
939c20789501 Uploaded
rico
parents:
diff changeset
1 #!/usr/bin/env python2.5
939c20789501 Uploaded
rico
parents:
diff changeset
2
939c20789501 Uploaded
rico
parents:
diff changeset
3 import os
939c20789501 Uploaded
rico
parents:
diff changeset
4 import sys
939c20789501 Uploaded
rico
parents:
diff changeset
5 from optparse import OptionParser
939c20789501 Uploaded
rico
parents:
diff changeset
6 import genome_diversity as gd
939c20789501 Uploaded
rico
parents:
diff changeset
7
939c20789501 Uploaded
rico
parents:
diff changeset
8 def main_function( parse_arguments=None ):
939c20789501 Uploaded
rico
parents:
diff changeset
9 if parse_arguments is None:
939c20789501 Uploaded
rico
parents:
diff changeset
10 parse_arguments = lambda arguments: ( None, arguments )
939c20789501 Uploaded
rico
parents:
diff changeset
11 def main_decorator( to_decorate ):
939c20789501 Uploaded
rico
parents:
diff changeset
12 def decorated_main( arguments=None ):
939c20789501 Uploaded
rico
parents:
diff changeset
13 if arguments is None:
939c20789501 Uploaded
rico
parents:
diff changeset
14 arguments = sys.argv
939c20789501 Uploaded
rico
parents:
diff changeset
15 options, arguments = parse_arguments( arguments )
939c20789501 Uploaded
rico
parents:
diff changeset
16 rc = 1
939c20789501 Uploaded
rico
parents:
diff changeset
17 try:
939c20789501 Uploaded
rico
parents:
diff changeset
18 rc = to_decorate( options, arguments )
939c20789501 Uploaded
rico
parents:
diff changeset
19 except Exception, err:
939c20789501 Uploaded
rico
parents:
diff changeset
20 sys.stderr.write( 'ERROR: %s\n' % str( err ) )
939c20789501 Uploaded
rico
parents:
diff changeset
21 traceback.print_exc()
939c20789501 Uploaded
rico
parents:
diff changeset
22 finally:
939c20789501 Uploaded
rico
parents:
diff changeset
23 sys.exit( rc )
939c20789501 Uploaded
rico
parents:
diff changeset
24 return decorated_main
939c20789501 Uploaded
rico
parents:
diff changeset
25 return main_decorator
939c20789501 Uploaded
rico
parents:
diff changeset
26
939c20789501 Uploaded
rico
parents:
diff changeset
27 def parse_arguments( arguments ):
939c20789501 Uploaded
rico
parents:
diff changeset
28 parser = OptionParser()
939c20789501 Uploaded
rico
parents:
diff changeset
29 parser.add_option('--input',
939c20789501 Uploaded
rico
parents:
diff changeset
30 type='string', dest='input',
939c20789501 Uploaded
rico
parents:
diff changeset
31 help='file of selected SNPs')
939c20789501 Uploaded
rico
parents:
diff changeset
32 parser.add_option('--output',
939c20789501 Uploaded
rico
parents:
diff changeset
33 type='string', dest='output',
939c20789501 Uploaded
rico
parents:
diff changeset
34 help='output file')
939c20789501 Uploaded
rico
parents:
diff changeset
35 parser.add_option('--primers_loc',
939c20789501 Uploaded
rico
parents:
diff changeset
36 type='string', dest='primers_loc',
939c20789501 Uploaded
rico
parents:
diff changeset
37 help='primers .loc file')
939c20789501 Uploaded
rico
parents:
diff changeset
38 parser.add_option('--scaffold_col',
939c20789501 Uploaded
rico
parents:
diff changeset
39 type="int", dest='scaffold_col',
939c20789501 Uploaded
rico
parents:
diff changeset
40 help='scaffold column in the input file')
939c20789501 Uploaded
rico
parents:
diff changeset
41 parser.add_option('--pos_col',
939c20789501 Uploaded
rico
parents:
diff changeset
42 type="int", dest='pos_col',
939c20789501 Uploaded
rico
parents:
diff changeset
43 help='position column in the input file')
939c20789501 Uploaded
rico
parents:
diff changeset
44 parser.add_option('--enzyme_list',
939c20789501 Uploaded
rico
parents:
diff changeset
45 type="string", dest='enzyme_list_string',
939c20789501 Uploaded
rico
parents:
diff changeset
46 help='comma separated list of enzymes')
939c20789501 Uploaded
rico
parents:
diff changeset
47 parser.add_option('--species',
939c20789501 Uploaded
rico
parents:
diff changeset
48 type="string", dest='species',
939c20789501 Uploaded
rico
parents:
diff changeset
49 help='species')
939c20789501 Uploaded
rico
parents:
diff changeset
50 return parser.parse_args( arguments[1:] )
939c20789501 Uploaded
rico
parents:
diff changeset
51
939c20789501 Uploaded
rico
parents:
diff changeset
52
939c20789501 Uploaded
rico
parents:
diff changeset
53 @main_function( parse_arguments )
939c20789501 Uploaded
rico
parents:
diff changeset
54 def main( options, arguments ):
939c20789501 Uploaded
rico
parents:
diff changeset
55 if not options.input:
939c20789501 Uploaded
rico
parents:
diff changeset
56 raise RuntimeError( 'missing --input option' )
939c20789501 Uploaded
rico
parents:
diff changeset
57 if not options.output:
939c20789501 Uploaded
rico
parents:
diff changeset
58 raise RuntimeError( 'missing --output option' )
939c20789501 Uploaded
rico
parents:
diff changeset
59 if not options.primers_loc:
939c20789501 Uploaded
rico
parents:
diff changeset
60 raise RuntimeError( 'missing --primers_loc option' )
939c20789501 Uploaded
rico
parents:
diff changeset
61 if not options.scaffold_col:
939c20789501 Uploaded
rico
parents:
diff changeset
62 raise RuntimeError( 'missing --scaffold_col option' )
939c20789501 Uploaded
rico
parents:
diff changeset
63 if not options.pos_col:
939c20789501 Uploaded
rico
parents:
diff changeset
64 raise RuntimeError( 'missing --pos_col option' )
939c20789501 Uploaded
rico
parents:
diff changeset
65 if not options.enzyme_list_string:
939c20789501 Uploaded
rico
parents:
diff changeset
66 raise RuntimeError( 'missing --enzyme_list option' )
939c20789501 Uploaded
rico
parents:
diff changeset
67 if not options.species:
939c20789501 Uploaded
rico
parents:
diff changeset
68 raise RuntimeError( 'missing --species option' )
939c20789501 Uploaded
rico
parents:
diff changeset
69
939c20789501 Uploaded
rico
parents:
diff changeset
70 snps = gd.SnpFile( filename=options.input, seq_col=int( options.scaffold_col ), pos_col=int( options.pos_col ) )
939c20789501 Uploaded
rico
parents:
diff changeset
71
939c20789501 Uploaded
rico
parents:
diff changeset
72 out_fh = gd._openfile( options.output, 'w' )
939c20789501 Uploaded
rico
parents:
diff changeset
73
939c20789501 Uploaded
rico
parents:
diff changeset
74 enzyme_dict = {}
939c20789501 Uploaded
rico
parents:
diff changeset
75 for enzyme in options.enzyme_list_string.split( ',' ):
939c20789501 Uploaded
rico
parents:
diff changeset
76 enzyme = enzyme.strip()
939c20789501 Uploaded
rico
parents:
diff changeset
77 if enzyme:
939c20789501 Uploaded
rico
parents:
diff changeset
78 enzyme_dict[enzyme] = 1
939c20789501 Uploaded
rico
parents:
diff changeset
79
939c20789501 Uploaded
rico
parents:
diff changeset
80 primer_data_file = gd.get_filename_from_loc( options.species, options.primers_loc )
939c20789501 Uploaded
rico
parents:
diff changeset
81 file_root, file_ext = os.path.splitext( primer_data_file )
939c20789501 Uploaded
rico
parents:
diff changeset
82 primer_index_file = file_root + ".cdb"
939c20789501 Uploaded
rico
parents:
diff changeset
83 primers = gd.PrimersFile( data_file=primer_data_file, index_file=primer_index_file )
939c20789501 Uploaded
rico
parents:
diff changeset
84
939c20789501 Uploaded
rico
parents:
diff changeset
85 comments_printed = False
939c20789501 Uploaded
rico
parents:
diff changeset
86
939c20789501 Uploaded
rico
parents:
diff changeset
87 while snps.next():
939c20789501 Uploaded
rico
parents:
diff changeset
88 seq, pos = snps.get_seq_pos()
939c20789501 Uploaded
rico
parents:
diff changeset
89 enzyme_list = primers.get_enzymes( seq, pos )
939c20789501 Uploaded
rico
parents:
diff changeset
90 for enzyme in enzyme_list:
939c20789501 Uploaded
rico
parents:
diff changeset
91 if enzyme in enzyme_dict:
939c20789501 Uploaded
rico
parents:
diff changeset
92 if not comments_printed:
939c20789501 Uploaded
rico
parents:
diff changeset
93 for comment in snps.comments:
939c20789501 Uploaded
rico
parents:
diff changeset
94 out_fh.write( "%s\n" % comment )
939c20789501 Uploaded
rico
parents:
diff changeset
95 comments_printed = True
939c20789501 Uploaded
rico
parents:
diff changeset
96 out_fh.write( "%s\n" % snps.line )
939c20789501 Uploaded
rico
parents:
diff changeset
97 break
939c20789501 Uploaded
rico
parents:
diff changeset
98
939c20789501 Uploaded
rico
parents:
diff changeset
99 out_fh.close()
939c20789501 Uploaded
rico
parents:
diff changeset
100
939c20789501 Uploaded
rico
parents:
diff changeset
101 if __name__ == "__main__":
939c20789501 Uploaded
rico
parents:
diff changeset
102 main()
939c20789501 Uploaded
rico
parents:
diff changeset
103