if __name__ == "__main__":
vcf_loc, out_loc = sys.argv[1], sys.argv[2]
inWindow = []
outFile = open(out_loc, "w")
# CSV output file format
# start, stop, #Alt sites, #Ref sites, Average depth, Alt Reads (sum from all samples), ref reads (sum from all samples),
# pseudo0:1000, pseudo0:1050, 5, 45, 197
#
outFile.write(
"start_chrom,start_pos,stop_chrom,stop_pos,altSiteCount,refSiteCount,avgDepth,avgRefReads,avgAltReads,avgOtherReads"
+ "\n")
with open(vcf_loc) as f:
for raw_line in f:
if len(inWindow) == WINDOW_SIZE:
#Analyze site
processWindow(inWindow, outFile)
inWindow = []
line = vcfLine(raw_line)
if line.isDataLine:
#Only add dataLines to the window
inWindow.append(line)
#process last incomplete block
processWindow(inWindow, outFile)
outFile.close()
import sys
from vcfLine import vcfLine
if __name__ == "__main__":
vcf_in,out_loc = sys.argv[1],sys.argv[2]
outFile = open(out_loc,"w")
with open(vcf_in) as vcfFile:
for line in vcfFile:
vcf_line = vcfLine(line)
if vcf_line.isDataLine:
outFile.write(vcf_line.chrom + " " + vcf_line.pos + "\n")
vcfFile.close()
outFile.close()
from vcfDict import vcfDict
from vcfLine import vcfLine
from vcfSample import vcfSample
from mpileLine import mpileLine
if __name__ == "__main__":
mpile_in,vcf_in,csv_out = sys.argv[1],sys.argv[2],sys.argv[3]
#Load dictionary of sites in vcf_in file
vcfSites = vcfDict(vcf_in)
vcfSites.loadDict()
#Create output file for csv_out
outFile = open(csv_out,"w")
with open(mpile_in) as pileFile:
for line in pileFile:
try:
pile_line = mpileLine(line)
print pile_line.repr()
if vcfSites.siteExists(pile_line.siteID):
vcf_line = vcfLine(vcfSites.getLine(pile_line.siteID))
except Exception, e:
print sys.exc_info()[0]
print traceback.format_exc()
print line
outFile.close()
pileFile.close()
import sys
from vcfLine import vcfLine
from vcfSample import vcfSample
HET = "0/1"
def allHet(samples):
"""
Determines if the list of samples is all HET samples
"""
for sample in samples:
if sample.GT != HET:
return False
return True
if __name__ == "__main__":
vcf_in, vcf_out = sys.argv[1], sys.argv[2]
outFile = open(vcf_out, "w")
with open(vcf_in) as file:
for line in file:
vLine = vcfLine(line)
if vLine.isDataLine:
if allHet(vLine.samples):
outFile.write(line)
outFile.close()
vcf_unfilt, vcf_targetSites, out_loc = sys.argv[1], sys.argv[2], sys.argv[
3]
outFile = open(out_loc, "w")
outFile.write(
"siteID,start_chrom,start_pos,stop_chrom,stop_pos,altSiteCount,refSiteCount,avgDepth,avgRefReads,avgAltReads,avgOtherReads"
+ "\n")
targetSitesDict = vcfDict(vcf_targetSites)
targetSitesDict.loadDict()
window = []
with open(vcf_unfilt) as unfilt:
for line in unfilt:
lineVcf = vcfLine(line)
if lineVcf.isDataLine: #Add data lines from unfiltered file to window
window.append(lineVcf)
if len(
window
) == WINDOW_SIZE + 1: #Remove oldest site seen keep size == WINDOW_SIZE
window.pop(0)
#Check if the site in the middle of the window is a target site
if checkSite(window[WINDOW_SIZE / 2], targetSitesDict):
processWindow(window, outFile)
outFile.close()
unfilt.close()
