We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

Variant analysis on 1 or 2 samples: should I skip the final steps?

kennethcondonkennethcondon Member
edited June 2018 in Ask the GATK team

Hello all,

I'm a bit confused as to what steps are necessary, and what steps are not going to add much benefit. I have 2 jobs to complete for 2 different research groups we support: 1) Germline short variant discovery on whole exome sequencing (WES) data collected from 1 mouse (1 sample in total), and 2) Germline short variant discovery on whole genome sequencing data (WGS) collected from 2 macaques (2 samples in total).

I have written a wrapper that follows the GATK best practices from fastq preprocessing to HaplotypeCaller with appropriate conditional loops and required files specific to each species and type of sequencing.

According to the GATK workflow - my next steps after running HaplotypeCaller (with --emit-ref-confidence GVCF) in the pipeline are 1) consolidate GVCFs, 2) Joint-calling cohort, and 3) VQSR).

So here are my concerns:

  1. Considering I have only 1 or 2 samples - is it pointless doing some/all of these steps? Should I just stick to the variants called in each sample by HaplotypeCaller? Should I remove "--emit-ref-confidence GVCF" and just create a regular VCF? Is it possible to hard filter a regular VCF?
  2. If I do VQSR, I don't know where to find truth sets. Someone has suggested to me that I can use the human truth set in other species because the information taken from the truth set is the profile of what a true SNP looks like, not the position of the SNP - I'm really not sure about this.

I have previously posted this on biostars without success.

Help Me, Obi-Wan.

Thanks in advance.


Sign In or Register to comment.