vcfped.py 文件源码

def bestGenders(genderdata):
    best = []
    sortfun = lambda x: (round(x['Xhetp'], 1), x['percentile'])
    for k,v in itertools.groupby(genderdata, key=lambda x: x['sample']):
        allcalls = list(v)
        noNA = [r for r in allcalls if r['gender'] != 'na']
        if not noNA: 
            best.append(allcalls[0])
            continue
        verdicts = {r['gender'] for r in noNA}
        if 'Male' in verdicts and 'Female' in verdicts:
            # if both genders are called, choose the last one
            noQ = [r for r in noNA if r['gender'] != '?']
            best.append(noQ[-1])
            continue
        if verdicts == {'?'}:
            verd = '?'
        else:
            verd = list(verdicts.difference({'?'}))[0]
        use = sorted([r for r in noNA if r['gender'] == verd], key=sortfun)
        best.append(use[0])

    return gender_table(best)