generate more candidates in the first step

[Luchixiang]

I've tried generating the candidates vcf file from HG001-NA12878-50x.bam of precisionFDA challenge by your code: DL4VC/tools/candidate_generator.py. We follwed the step-by-step tutorials. However, the generated candidates vcf file by us has about 16000,000 records and the candidates file provided by you on github contains only 9600,000 records. I'm wondering why we generated more candidates as we didn't modify the code and used the same threshhold as yours(0.075 for snp_min_freq and 0.02 for indel_min_freq). Can you please help me to figure out the reason?

[quanguo2088]

你好，请问一下你还在做变异检测嘛，我想问一下DL4VC训练大概要多久呀，有没有大概的时间呢，我估算了一下我的一个epoch要好几天。。