def getLocusAndData(self, inputFname, VCFOutputType=2):
"""
2011-9-21
"""
contig_id_pattern = re.compile(r'Contig(\d+).*')
contig2locus2frequency = {}
fname = inputFname
if fname[-6:]!='vcf.gz' and fname[-3:]!='vcf':
return None
sys.stderr.write("%s ..."%fname)
contig_id_pattern_sr = contig_id_pattern.search(inputFname)
if contig_id_pattern_sr:
contig_id = contig_id_pattern_sr.group(1)
else:
contig_id = os.path.splitext(os.path.split(inputFname)[1])[0]
vcfFile = VCFFile(inputFname=self.inputFname)
counter = 0
real_counter = 0
locus_ls = []
xData_ls = []
yData_ls = []
for vcfRecord in vcfFile.parseIter():
locus_id = vcfRecord.locus_id
chr = vcfRecord.chr
pos = vcfRecord.pos
pos = int(pos)
AF1 = vcfRecord.info_tag2value.get("AF", vcfRecord.info_tag2value.get("AF1", None))
if AF1:
AF1 = float(AF1)
locus_ls.append(locus_id)
xData_ls.append(pos)
yData_ls.append(AF1)
sys.stderr.write("%s loci. Done.\n"%(len(yData_ls)))
return PassingData(contig_id=contig_id, locus_ls=locus_ls, yData_ls=yData_ls, xData_ls=xData_ls)
