stats in chinese from 1000 genome

1000.py

#coding=utf-8
import sys,json,re
import vcf

vcf_reader = vcf.Reader(open(sys.argv[1]))
ch = json.load(open('chb.json')) + json.load(open('chs.json'))
pattern = r'^([0,1])[\|\/]([0,1])$'

for record in vcf_reader:
    if record.var_type == 'snp' and len(record.ALT) == 1:
        stats = {0:0, 1:0, 2:0}
        for s in record.samples:
            if s.sample in ch and re.match(pattern, s.gt_nums):
                g = sum([int(i) for i in re.match(pattern, s.gt_nums).groups()])
                stats[g] = stats[g] + 1
        print record.ID, record.REF, record.ALT[0], stats[0], stats[1], stats[2]