39
|
1 #!/usr/bin/env python
|
|
2
|
|
3 import sys,string
|
|
4 sys.path.insert(0,"../CGDataNew")
|
|
5
|
|
6 import CGData.GenomicSegment
|
|
7 import CGData.SegToMatrix
|
|
8 import CGData.RefGene
|
|
9 import CGData.GeneMap
|
|
10 import optparse
|
|
11
|
|
12 class segs:
|
|
13 def __init__(self):
|
|
14 self.probes = []
|
|
15
|
|
16 def load (self, handle): #handle bed6
|
|
17 fin =open(handle,'r')
|
|
18 while 1:
|
|
19 line =string.strip(fin.readline())
|
|
20 if line =="":
|
|
21 break
|
|
22 if line[0]=="#":
|
|
23 continue
|
|
24 tmp = string.split(line,"\t")
|
|
25 if len(tmp)== 5:
|
|
26 p = probeseg(tmp[0], tmp[1], int(tmp[3]), int(tmp[4]),tmp[2])
|
|
27 self.probes.append(p)
|
|
28 fin.close()
|
|
29
|
|
30
|
|
31 class probeseg:
|
|
32 def __init__(self, name, chrom, chrom_start, chrom_end, strand):
|
|
33 self.name = name
|
58
|
34 if chrom[0:3] in ["chr","CHR","Chr"]:
|
|
35 chrom = "chr"+chrom[3:]
|
|
36 else:
|
|
37 chrom = "chr"+chrom
|
39
|
38 self.chrom = chrom
|
|
39 self.chrom_start = chrom_start
|
|
40 self.chrom_end = chrom_end
|
|
41 self.strand = strand
|
|
42
|
|
43
|
|
44 if __name__ == "__main__":
|
|
45 def printUsage():
|
|
46 print "python segToProbeMap.py segInput(name,chr,strand,start,end) refGene(eg hg18) probeMapOut --mode=cnv|exp\n"
|
|
47
|
|
48 if len(sys.argv) != 5:
|
|
49 printUsage()
|
|
50 sys.exit()
|
|
51
|
|
52 parser = optparse.OptionParser()
|
|
53 parser.add_option("--mode", action="store", type="string", dest="mode")
|
|
54 (options, args) = parser.parse_args()
|
|
55
|
|
56 if (not options.mode) or (options.mode not in ["cnv","exp"]):
|
|
57 printUsage()
|
|
58 sys.exit()
|
|
59
|
|
60 probes=segs()
|
|
61 probes.load(sys.argv[1])
|
|
62
|
|
63 refgene = CGData.RefGene.RefGene()
|
|
64 refgene.load(sys.argv[2])
|
|
65
|
|
66 handle = open(sys.argv[3], "w")
|
|
67 if options.mode=="cnv":
|
|
68 probeMapper = CGData.GeneMap.ProbeMapper('b')
|
|
69 if options.mode=="exp":
|
|
70 probeMapper = CGData.GeneMap.ProbeMapper('g')
|
|
71
|
|
72 handle.write("%s\t%s\t%s\t%s\t%s\t%s\n" % ("#id", "gene","chrom","chromStart","chromEnd","strand"))
|
|
73 for probe in probes.probes:
|
|
74 hits = []
|
|
75 for hit in probeMapper.find_overlap( probe, refgene ):
|
|
76 if hit.name not in hits:
|
|
77 hits.append(hit.name)
|
|
78 handle.write("%s\t%s\t%s\t%s\t%s\t%s\n" % (probe.name, ",".join(hits), probe.chrom, probe.chrom_start, probe.chrom_end, probe.strand))
|
|
79 handle.close()
|
|
80
|
|
81
|