Where do I find resources for using FilterVariantTranches with CNN_2D/CNN_1D scores?

Answered

Follow

Rodrigo Cabrera

• July 29, 2021 20:00

Documentation for FilterVariantTranches suggests using hapmap.vcf and mills.vcf resources to select cutoffs for CNN_2D and CNN_1D scores:

 gatk FilterVariantTranches \
   -V input.vcf.gz \
   --resource hapmap.vcf \
   --resource mills.vcf \
   --info-key CNN_2D \
   --snp-tranche 99.95 \
   --indel-tranche 99.4 \
   --invalidate-previous-filters \
   -O filtered.vcf

I imagine these files need CNN_2D or CNN_1D scores. The "resources_broad_hg38_v0_Mills_and_1000G_gold_standard.indels.hg38" and "resources_broad_hg38_v0_hapmap_3.3" files are not scored for CNN_2D or CNN_1D so I assume they will not work.

Could you provide information on where to find these resources or other ways to run this tool?

Thanks!

0

• 

  Genevieve Brandt (she/her)

  • August 07, 2021 00:18

  Hi Rodrigo Cabrera,

  The files don't need to be scored for CNN_1D or CNN_2D to be used as resources with FilterVariantTranches. They are just supposed to be known sites files.

  Best,

  Genevieve

  0

  Comment actions Permalink

Please sign in to leave a comment.