Mercurial > repos > drosofff > lumpy

annotate extractSplitReads_BwaMem.py @ 0:8b3daa745d9b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
author drosofff
date Tue, 06 Dec 2016 05:46:28 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
1 #!/usr/bin/env python
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
2
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
3 import sys
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
4 import getopt
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
5 import string
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
6 from optparse import OptionParser
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
7 import re
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
8
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
9 def extractSplitsFromBwaMem(inFile,numSplits,includeDups,minNonOverlap):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
10 if inFile == "stdin":
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
11 data = sys.stdin
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
12 else:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
13 data = open(inFile, 'r')
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
14 for line in data:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
15 split = 0
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
16 if line[0] == '@':
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
17 print line.strip()
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
18 continue
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
19 samList = line.strip().split('\t')
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
20 sam = SAM(samList)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
21 if includeDups==0 and (1024 & sam.flag)==1024:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
22 continue
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
23 for el in sam.tags:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
24 if "SA:" in el:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
25 if(len(el.split(";")))<=numSplits:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
26 split = 1
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
27 mate = el.split(",")
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
28 mateCigar = mate[3]
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
29 mateFlag = int(0)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
30 if mate[2]=="-": mateFlag = int(16)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
31 if split:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
32 read1 = sam.flag & 64
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
33 if read1 == 64: tag = "_1"
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
34 else: tag="_2"
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
35 samList[0] = sam.query + tag
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
36 readCigar = sam.cigar
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
37 readCigarOps = extractCigarOps(readCigar,sam.flag)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
38 readQueryPos = calcQueryPosFromCigar(readCigarOps)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
39 mateCigarOps = extractCigarOps(mateCigar,mateFlag)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
40 mateQueryPos = calcQueryPosFromCigar(mateCigarOps)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
41 overlap = calcQueryOverlap(readQueryPos.qsPos,readQueryPos.qePos,mateQueryPos.qsPos,mateQueryPos.qePos)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
42 nonOverlap1 = 1 + readQueryPos.qePos - readQueryPos.qsPos - overlap
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
43 nonOverlap2 = 1 + mateQueryPos.qePos - mateQueryPos.qsPos - overlap
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
44 mno = min(nonOverlap1, nonOverlap2)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
45 if mno >= minNonOverlap:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
46 print "\t".join(samList)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
47
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
48 #--------------------------------------------------------------------------------------------------
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
49 # functions
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
50 #--------------------------------------------------------------------------------------------------
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
51
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
52 class SAM (object):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
53 """
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
54 __very__ basic class for SAM input.
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
55 """
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
56 def __init__(self, samList = []):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
57 if len(samList) > 0:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
58 self.query = samList[0]
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
59 self.flag = int(samList[1])
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
60 self.ref = samList[2]
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
61 self.pos = int(samList[3])
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
62 self.mapq = int(samList[4])
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
63 self.cigar = samList[5]
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
64 self.matRef = samList[6]
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
65 self.matePos = int(samList[7])
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
66 self.iSize = int(samList[8])
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
67 self.seq = samList[9]
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
68 self.qual = samList[10]
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
69 self.tags = samList[11:]#tags is a list of each tag:vtype:value sets
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
70 self.valid = 1
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
71 else:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
72 self.valid = 0
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
73 self.query = 'null'
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
74
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
75 def extractTagValue (self, tagID):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
76 for tag in self.tags:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
77 tagParts = tag.split(':', 2);
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
78 if (tagParts[0] == tagID):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
79 if (tagParts[1] == 'i'):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
80 return int(tagParts[2]);
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
81 elif (tagParts[1] == 'H'):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
82 return int(tagParts[2],16);
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
83 return tagParts[2];
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
84 return None;
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
85
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
86 #-----------------------------------------------
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
87 cigarPattern = '([0-9]+[MIDNSHP])'
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
88 cigarSearch = re.compile(cigarPattern)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
89 atomicCigarPattern = '([0-9]+)([MIDNSHP])'
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
90 atomicCigarSearch = re.compile(atomicCigarPattern)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
91
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
92 def extractCigarOps(cigar,flag):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
93 if (cigar == "*"):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
94 cigarOps = []
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
95 elif (flag & 0x0010):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
96 cigarOpStrings = cigarSearch.findall(cigar)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
97 cigarOps = []
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
98 for opString in cigarOpStrings:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
99 cigarOpList = atomicCigarSearch.findall(opString)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
100 # print cigarOpList
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
101 # "struct" for the op and it's length
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
102 cigar = cigarOp(cigarOpList[0][0], cigarOpList[0][1])
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
103 # add to the list of cigarOps
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
104 cigarOps.append(cigar)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
105 cigarOps = cigarOps
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
106 cigarOps.reverse()
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
107 ##do in reverse order because negative strand##
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
108 else:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
109 cigarOpStrings = cigarSearch.findall(cigar)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
110 cigarOps = []
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
111 for opString in cigarOpStrings:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
112 cigarOpList = atomicCigarSearch.findall(opString)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
113 # "struct" for the op and it's length
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
114 cigar = cigarOp(cigarOpList[0][0], cigarOpList[0][1])
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
115 # add to the list of cigarOps
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
116 cigarOps.append(cigar)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
117 # cigarOps = cigarOps
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
118 return(cigarOps)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
119
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
120 def calcQueryPosFromCigar(cigarOps):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
121 qsPos = 0
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
122 qePos = 0
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
123 qLen = 0
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
124 # if first op is a H, need to shift start position
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
125 # the opPosition counter sees if the for loop is looking at the first index of the cigar object
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
126 opPosition = 0
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
127 for cigar in cigarOps:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
128 if opPosition == 0 and (cigar.op == 'H' or cigar.op == 'S'):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
129 qsPos += cigar.length
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
130 qePos += cigar.length
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
131 qLen += cigar.length
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
132 elif opPosition > 0 and (cigar.op == 'H' or cigar.op == 'S'):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
133 qLen += cigar.length
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
134 elif cigar.op == 'M' or cigar.op == 'I':
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
135 qePos += cigar.length
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
136 qLen += cigar.length
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
137 opPosition += 1
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
138 d = queryPos(qsPos, qePos, qLen);
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
139 return d
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
140
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
141 class cigarOp (object):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
142 """
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
143 sturct to store a discrete CIGAR operations
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
144 """
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
145 def __init__(self, opLength, op):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
146 self.length = int(opLength)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
147 self.op = op
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
148
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
149 class queryPos (object):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
150 """
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
151 struct to store the start and end positions of query CIGAR operations
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
152 """
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
153 def __init__(self, qsPos, qePos, qLen):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
154 self.qsPos = int(qsPos)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
155 self.qePos = int(qePos)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
156 self.qLen = int(qLen)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
157
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
158
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
159 def calcQueryOverlap(s1,e1,s2,e2):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
160 o = 1 + min(e1, e2) - max(s1, s2)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
161 return max(0, o)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
162
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
163 ###############################################
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
164
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
165 class Usage(Exception):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
166 def __init__(self, msg):
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
167 self.msg = msg
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
168
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
169 def main():
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
170
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
171 usage = """%prog -i <file>
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
172
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
173 extractSplitReads_BwaMem v0.1.0
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
174 Author: Ira Hall
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
175 Description: Get split-read alignments from bwa-mem in lumpy compatible format. Ignores reads marked as duplicates.
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
176 Works on read or position sorted SAM input. Tested on bwa mem v0.7.5a-r405.
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
177 """
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
178 parser = OptionParser(usage)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
179
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
180 parser.add_option("-i", "--inFile", dest="inFile",
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
181 help="A SAM file or standard input (-i stdin).",
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
182 metavar="FILE")
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
183 parser.add_option("-n", "--numSplits", dest="numSplits", default=2, type = "int",
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
184 help="The maximum number of split-read mappings to allow per read. Reads with more are excluded. Default=2",
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
185 metavar="INT")
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
186 parser.add_option("-d", "--includeDups", dest="includeDups", action="store_true",default=0,
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
187 help="Include alignments marked as duplicates. Default=False")
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
188 parser.add_option("-m", "--minNonOverlap", dest="minNonOverlap", default=20, type = "int",
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
189 help="minimum non-overlap between split alignments on the query (default=20)",
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
190 metavar="INT")
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
191 (opts, args) = parser.parse_args()
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
192 if opts.inFile is None:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
193 parser.print_help()
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
194 print
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
195 else:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
196 try:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
197 extractSplitsFromBwaMem(opts.inFile, opts.numSplits, opts.includeDups, opts.minNonOverlap)
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
198 except IOError as err:
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
199 sys.stderr.write("IOError " + str(err) + "\n");
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
200 return
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
201 if __name__ == "__main__":
8b3daa745d9b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/lumpy commit c0bfc4b2215705e1b5fd1d4e60b1d72e5da13c92
drosofff
parents:
diff changeset
202 sys.exit(main())