Hi,
I'm looking to use GATK's independent variant calling + joint genotyping workflow on a large pedigree with 10 different families. Is there any difference between the two approaches below:
I ask this because if the result of genotyping one family does not affect the results for other families, I might as well run them in parallel.
What do you say?
EDIT: I'm closing this question because I was mistaken in my assumption that GATK uses pedigree information during variant calling (HaplotypeCaller) or genotyping (GenotypeGVCFs).
I am no expert on GATK, but I believe that it does not incorporate pedigree information during the variant calling. Calling all the samples simultaneously will give somewhat better results for shared variants, although sensitivity to singletons may be reduced somewhat, so it depends a little on what your goals are. Both of these effects will be more pronounced in borderline cases (low coverage).
You might also want to consider using a variant caller such as Real Time Genomics that can directly incorporate pedigree information for multiple families during the calling (and give you a head start on de novo identification).
Thank you for the inputs - I did not state my question clearly and I apologize for that. I was comparing running the entire pipeline (align+recalibrate+call+genotype) for all samples simultaneously versus per family. Does GATK incorporate pedigree information during genotyping?
And I've never heard of RTG - thank you for the pointer!
Log in to answer this question.
