How to interpret results from GATK ApplyVQSR

Question

[kittys]

Hello

I am currently using the GATK v4 best practises pipeline for short germline variant calling. The final step is applying a variant quality recalibration model to the output vcf file. However the documentation for this is absolutely awful and I am struggling to identify which variants are the ones I want to keep - are the ones labelled as PASS ones I want to select? Or the ones that contain tranche values?

The only clue I can find in old documentation is that ones labelled as "filtered" should be discarded - but this word doesn't appear anyway in my output file.

Any help is massively appreciated!!

Answer 1

[Nicolas Rosewick]

Did you look at the tranche plot generated at the VariantRecalibrator step ? You should then use an appropriate thresold based on this tranche plot.

See : https://software.broadinstitute.org/gatk/documentation/article?id=23216 and here : https://software.broadinstitute.org/gatk/documentation/article.php?id=1259

after ApplyVQSR, only "PASS" variants should be used for further analysis