Does Anybody Know Which Gatk Walkers Are Positively Affected By Pedigree Information?

Question

[Matthieu Miossec]

Hello,

I am calling variants in a set of trios and I am trying to obtain the best results possible from GATK. It has been my observation that adding pedigree information to the last phase of GATK, as suggested by the 'Best Practices' infographic doesn't seem to actually change my final results. Elsewhere on the website, it is suggested to pass GATK to all walkers and see what comes out of it. I'd like to know if anyone has looked into which walkers actually derive some positive effect from pedigree files (apart from PhaseByTransmission). The GATK website doesn't provide such answers as far as I've seen.

One thing I've considered is that my pedigree file doesn't contain genotype information and that might be something that the last phase of GATK uses (in the same sense that it uses known genotypes from other sources). Has anybody observed just that?

Answer 1

[vdauwera]

You're right that few of the GATK walkers actually use pedigree information, aside from the obvious PhaseByTransmission walker. The main reason is that analyses downstream of having generated the high quality callset, where pedigree information is most meaningful, aren't really within the scope of our activity. IIRC the callers can use pedigree information when computing AlleleFrequency and InbreedingCoefficient annotations. VariantAnnotator can use the ped file if you ask it to annotate things like transmission disequilibrium or mendelian violations.