Bug Bulletin: we have identified a bug that affects indexing when producing gzipped VCFs. This will be fixed in the upcoming 3.2 release; in the meantime you need to reindex gzipped VCFs using Tabix.

HaplotypeCaller to call variants common between samples

FarahFarah Posts: 8Member

Hi,

I'm using HaplotypeCaller on a family that is unusual; it has two affected children and one unaffected parent. The disease is expected to be caused by a homozygous or compound heterozygous mutation, and the unaffected parent is a carrier. We don't have any sequence data on the other unaffected parent. So obviously the variants I am interested in are going to be common among all three samples, i.e., child1, child2 and parent. I ran HaplotypeCaller on each sample seperately. I now have 3 vcf files. How can I filter the common variants among them? I have added read groups that show the relationship with tags 'c1' 'c2' and 'p' (for child1, child2 and parent). **Or **should I go back and rerun HaplotypeCaller but this time input all three sample files together and specify one output vcf? I am reluctant to do this as it took about 8 days for each of my HaplotypeCaller runs to finish the last time. [plus I actually have 6 families to do this on, not one].

Please help. And appreciate your support

Answers

  • CarneiroCarneiro Posts: 271Administrator, GSA Member admin
  • FarahFarah Posts: 8Member

    Ok, I can use first use CombineVariants and then SelectVariants to filter out common lines. But I will have to do this twice right? As CombineVariants can only handle 2 input files. I am using SnpEff to annotate my vcf files. Can I use CombineVariants and then SelectVariants first to pull out the common lines after I add in the annotations or no?

    [I also found BEDtools has 'intersect' which can work like CombineVariants - SelectVariants - could I use that with GATK generated vcf files? - any preference]

    Thanks,

  • aeonsimaeonsim Posts: 45Member

    I would use vcftools', vcf-merge script to merge each of your family sets of VCF files to create a single VCF file per family then use SnpSift (from SNPEffect Predictor/snpEFF) to create a filter that selects variants that are present in all 3 individuals. ie "isVariant(GEN[0]) & isVariant(GEN[1]) & isVariant(Gen[2])" you can also filter on things like isHet, isHom, hetCount, homCount if your confidant of the Variant type.

    After you've merged with the vcftools script you may also be able to use GATK SelectVariants to filter however I am unfamiliar with the use of that tool.

Sign In or Register to comment.