Mercurial > repos > devteam > substitutions

comparison substitutions.py @ 0:190735ce4c2b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Imported from capsule None
author devteam
date Tue, 01 Apr 2014 09:12:31 -0400
parents
children 385a5c3d7244
comparison
equal deleted inserted replaced
-1:000000000000 0:190735ce4c2b
1 #!/usr/bin/env python
2 #Guruprasad ANanda
3 """
4 Fetches substitutions from pairwise alignments.
5 """
6
7 from galaxy import eggs
8
9 from galaxy.tools.util import maf_utilities
10
11 import bx.align.maf
12 import sys
13
14 def stop_err(msg):
15 sys.stderr.write(msg)
16 sys.exit()
17
18
19 if len(sys.argv) < 3:
20 stop_err("Incorrect number of arguments.")
21
22 inp_file = sys.argv[1]
23 out_file = sys.argv[2]
24 fout = open(out_file, 'w')
25
26 def fetchSubs(block):
27 src1 = block.components[0].src
28 sequence1 = block.components[0].text
29 start1 = block.components[0].start
30 end1 = block.components[0].end
31 len1_withgap = len(sequence1)
32
33 for seq in range (1, len(block.components)):
34 src2 = block.components[seq].src
35 sequence2 = block.components[seq].text
36 start2 = block.components[seq].start
37 end2 = block.components[seq].end
38 sub_begin = None
39 sub_end = None
40 begin = False
41
42 for nt in range(len1_withgap):
43 if sequence1[nt] not in '-#$^*?' and sequence2[nt] not in '-#$^*?': # Not a gap or masked character
44 if sequence1[nt].upper() != sequence2[nt].upper():
45 if not(begin):
46 sub_begin = nt
47 begin = True
48 sub_end = nt
49 else:
50 if begin:
51 print >> fout, "%s\t%s\t%s" % ( src1, start1+sub_begin-sequence1[0:sub_begin].count('-'), start1+sub_end-sequence1[0:sub_end].count('-') )
52 print >> fout, "%s\t%s\t%s" % ( src2, start2+sub_begin-sequence2[0:sub_begin].count('-'), start2+sub_end-sequence2[0:sub_end].count('-') )
53 begin = False
54 else:
55 if begin:
56 print >> fout, "%s\t%s\t%s" % ( src1, start1+sub_begin-sequence1[0:sub_begin].count('-'), end1+sub_end-sequence1[0:sub_end].count('-') )
57 print >> fout, "%s\t%s\t%s" % ( src2, start2+sub_begin-sequence2[0:sub_begin].count('-'), end2+sub_end-sequence2[0:sub_end].count('-') )
58 begin = False
59
60
61 def main():
62 skipped = 0
63 not_pairwise = 0
64 try:
65 maf_reader = bx.align.maf.Reader( open(inp_file, 'r') )
66 except:
67 stop_err("Your MAF file appears to be malformed.")
68 print >> fout, "#Chr\tStart\tEnd"
69 for block in maf_reader:
70 if len(block.components) != 2:
71 not_pairwise += 1
72 continue
73 try:
74 fetchSubs(block)
75 except:
76 skipped += 1
77
78 if not_pairwise:
79 print "Skipped %d non-pairwise blocks" % (not_pairwise)
80 if skipped:
81 print "Skipped %d blocks" % (skipped)
82
83
84 if __name__ == "__main__":
85 main()