annotate VCFGandalfTools/VCFFiltering_wrapper.py @ 2:6bebeb76fa8d draft

Uploaded
author urgi-team
date Tue, 05 Apr 2016 08:33:41 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
1 #!/usr/bin/env python
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
2
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
3
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
4 import subprocess, tempfile, sys, os, glob, shutil, time
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
5 from optparse import OptionParser
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
6 from optparse import Option, OptionValueError
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
7
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
8 class VCFFilteringWrapper(object):
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
9
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
10 def __init__(self):
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
11 self._options = None
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
12
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
13
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
14 def stop_err(self, msg ):
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
15 sys.stderr.write( "%s\n" % msg )
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
16 sys.exit()
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
17
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
18
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
19 def setAttributesFromCmdLine(self):
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
20 description = "VCFFiltering_wrapper"
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
21 description += "\nWrapper for VCFFiltering ;\n VCFFiltering filters SNP on a VCF depending on depth (DP) allele number (AN), allele frequency (AF) and SNP quality.\n"
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
22 description += "example 1 : VCFFiltering.py -f myVCF.vcf -o FilteredVCF.vcf\n"
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
23 description += "example 2 : VCFFiltering.py -f myVCF.vcf -N 2 -F 0.87 -b bed1.bed bed2.bed -o FilteredVCF.vcf\n"
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
24 parser = OptionParser(description = description, version = "0.2")
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
25 parser.add_option("-f", "--vcf", dest = "VCFFile", action = "store", type = "string", help = "Input VCF File name [compulsory] [format: VCF]", default = "")
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
26 parser.add_option("-o", "--output", dest = "outFile", action = "store", type = "string", help = "output VCF File name [compulsory] [format: VCF]", default = "")
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
27 parser.add_option("-m", "--minDP", dest = "minDP", action = "store", type = "int", help = "minimum of depth ; if both minDP and maxDP are set, optimal DP will not be calculated ", default = 0)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
28 parser.add_option("-M", "--maxDP", dest = "maxDP", action = "store", type = "int", help = "maximum of depth ; if both minDP and maxDP are set, optimal DP will not be calculated ", default = 0)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
29 parser.add_option("-N", "--AN", dest = "AN", action = "store", type = "int", help = "maximum number of allele for a SNP; default = 2", default = 2)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
30 parser.add_option("-F", "--AF", dest = "AF", action = "store", type = "float", help = "minimum frequency for the alternative allele of a SNP; default = 0.9", default = 0.9)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
31 parser.add_option("-b", "--bed", dest = "bedFiles", action = "append", type = "string", help = "bed files: list of coordinates to filter, multiple arguments allowed '-b file1 file2' ", default = [])
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
32 parser.add_option("-G", "--graphHTML", dest = "graphHTML", action = "store", type = "string", help = "name of the HTML linking to graphs ", default = "")
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
33 parser.add_option("-d", "--dirGraphs", dest = "dirGraphs", action = "store", type = "string", help = "name of the folder containing graphs ", default = "")
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
34 options = parser.parse_args()[0]
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
35 self._setAttributesFromOptions(options)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
36
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
37
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
38 def _setAttributesFromOptions(self, options):
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
39 self._options = options
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
40
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
41 def run(self):
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
42 if self._options.minDP and self._options.maxDP :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
43 if self._options.minDP > self._options.maxDP :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
44 self.stop_err( 'error in options : minDP > max DP (%s > %s)' %(self._options.minDP,self._options.maxDP))
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
45
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
46 prg = "VCFFiltering.py -g -G 'png' "
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
47 args = ""
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
48 args += "-f %s" % self._options.VCFFile
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
49 args += " "
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
50 args += "-o %s" % self._options.outFile
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
51 if self._options.AF :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
52 args += " "
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
53 args += "-F %s" % self._options.AF
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
54 if self._options.AN :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
55 args += " "
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
56 args += "-N %s" % self._options.AN
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
57 if self._options.minDP :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
58 args += " "
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
59 args += "-m %s" % self._options.minDP
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
60 if self._options.maxDP :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
61 args += " "
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
62 args += "-M %s" % self._options.maxDP
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
63 for bedfile in self._options.bedFiles :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
64 args += " "
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
65 args += "-b %s" % bedfile
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
66 cmd = "%s %s" %(prg, args)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
67
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
68 print cmd
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
69
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
70 try:
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
71 tmp_err = tempfile.NamedTemporaryFile().name
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
72 tmp_stderr = open( tmp_err, 'wb' )
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
73 proc = subprocess.Popen( args=cmd, shell=True, cwd=".", stderr=tmp_stderr )
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
74 returncode = proc.wait()
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
75 tmp_stderr.close()
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
76 # get stderr, allowing for case where it's very large
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
77 tmp_stderr = open( tmp_err, 'rb' )
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
78 stderr = ''
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
79 buffsize = 1048576
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
80 try:
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
81 while True:
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
82 stderr += tmp_stderr.read( buffsize )
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
83 if not stderr or len( stderr ) % buffsize != 0:
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
84 break
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
85 except OverflowError:
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
86 pass
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
87 tmp_stderr.close()
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
88 if stderr:
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
89 raise Exception, stderr
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
90 except Exception, e:
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
91 self.stop_err( 'Error in VCFFiltering:\n' + str( e ) )
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
92
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
93 if True :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
94 html = open(self._options.graphHTML, "w")
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
95
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
96 os.mkdir(self._options.dirGraphs)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
97 lGraphsFiles = glob.glob("VCFFiltering_graphs/*")
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
98 for file in lGraphsFiles :
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
99 baseName = os.path.basename(file)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
100 shutil.move( file ,"%s/%s" %(self._options.dirGraphs, baseName))
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
101 line = "<img src=\"%s\" > \n" %(baseName)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
102 html.write(line)
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
103
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
104
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
105 if __name__ == "__main__":
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
106 iWrapper = VCFFilteringWrapper()
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
107 iWrapper.setAttributesFromCmdLine()
6bebeb76fa8d Uploaded
urgi-team
parents:
diff changeset
108 iWrapper.run()