Hi all,

I called variant using Deepvariant (via clara https://catalog.ngc.nvidia.com/orgs/nvidia/teams/clara/containers/clara-parabricks-deepvariant) using the following command:

clara.sif \
  pbrun deepvariant  \
    --ref=GRCh38.fasta  \
    --in-bam=mysample_readgroup_sorted.bam \
    --out-variants=mysample.vcf \
    --num-cpu-threads-per-stream=6 \
    --num-streams-per-gpu=4d 

It worked fine however 50% of the variant didn't pass the filtering process. It is not a problem but I was wondering what is the default filtering that DeepVariant uses? In the github I found this "No filtering is needed beyond setting your preferred minimum quality threshold." but then I didn't specify any so why would half of it fail at the filtering step?

Thanks for your help