Hi all, I am trying to decide between joint and simple sample haplotypecaller. My goal is to get a list of all the common variants called across my samples and then filter them out of the vcfs. I have already run with simple sample. Will running with joint replace/delete the results from simple sample? What to take into account to decide between the 2?

I have already run with simple sample. Will running with joint replace/delete the results from simple sample? By default, yes, but If you change the output folder name (

--outdir ./results/run30Aug2024

, for example), it won’t, and then you could make a comparison?

Hi thanks for the response. Yes that would be one solution. Is that good practice? Could I keep the same work dir and skip the preprocessing part? Or use --step and use the files from the first run as input for the joint call?

I think yeah it is good to be in the habit of modifying the output folder, just to help you organise your batches - others may feel differently. I think if you use

-resume

it should even cache the steps that can be cached?

I do not know if running haplotypecaller with joint enabled counts as a different process ( like running freebayes for example) or not.

Different variant call parameters will for sure - but your alignment and such is the same and that should be cached. Suggest trying on a dummy example (one or two samples) to test?

So you think that running haplotypecaller with joint would create a different subfolder results/variant_calling/haplotypecaller_joint ? I will try to test 1 sample on a new result folder.

I would do a full re-run and change the

--outdir

parameter to point all the output somewhere else, while using the

-resume

flag to make it only re-run stuff that needs it.

I will do that. Thanks!

The pipeline will not use HaplotypeCaller for the joint variant calling. Each sample is still processed individually with HaplotypeCaller and the resulting gVCF files will be jointly genotyped by GENOTYPEGVCFS. The output of that step will be in

"${params.outdir}/variant_calling/haplotypecaller/joint_variant_calling

, so nothing will be overwritten.

Hello, What do you mean "The pipeline will not use HaplotypeCaller for the joint variant calling." I have run now for 1 sample and I can see how the output is distinct.