We’re moving the GATK website, docs and forum to a new platform. Read the full story and breakdown of key changes on this blog.
If you happen to see a question you know the answer to, please do chime in and help your fellow community members. We encourage our fourm members to be more involved, jump in and help out your fellow researchers with their questions. GATK forum is a community forum and helping each other with using GATK tools and research is the cornerstone of our success as a genomics research community.We appreciate your help!
Test-drive the GATK tools and Best Practices pipelines on Terra
Check out this blog post to learn how you can get started with GATK and try out the pipelines in preconfigured workspaces (with a user-friendly interface!) without having to install anything.
(howto) Run the genotype refinement workflow
This tutorial describes step-by-step instruction for applying the Genotype Refinement workflow (described in this method article) to your data.
Step 1: Derive posterior probabilities of genotypes
In this first step, we are deriving the posteriors of genotype calls in our callset,
recalibratedVariants.vcf, which just came out of the VQSR filtering step; it contains among other samples a trio of individuals (mother, father and child) whose family structure is described in the pedigree file
trio.ped (which you need to supply). To do this, we are using the most comprehensive set of high confidence SNPs available to us, a set of sites from Phase 3 of the 1000 Genomes project (available in our resource bundle), which we pass via the
java -jar GenomeAnalysisToolkit.jar -R human_g1k_v37_decoy.fasta -T CalculateGenotypePosteriors --supporting 1000G_phase3_v4_20130502.sites.vcf -ped trio.ped -V recalibratedVariants.vcf -o recalibratedVariants.postCGP.vcf
This produces the output file
recalibratedVariants.postCGP.vcf, in which the posteriors have been annotated wherever possible.
Step 2: Filter low quality genotypes
In this second, very simple step, we are tagging low quality genotypes so we know not to use them in our downstream analyses. We use Q20 as threshold for quality, which means that any passing genotype has a 99% chance of being correct.
java -jar $GATKjar -T VariantFiltration -R $bundlePath/b37/human_g1k_v37_decoy.fasta -V recalibratedVariants.postCGP.vcf -G_filter "GQ < 20.0" -G_filterName lowGQ -o recalibratedVariants.postCGP.Gfiltered.vcf
Note that in the resulting VCF, the genotypes that failed the filter are still present, but they are tagged
lowGQ with the FT tag of the FORMAT field.
Step 3: Annotate possible de novo mutations
In this third and final step, we tag variants for which at least one family in the callset shows evidence of a de novo mutation based on the genotypes of the family members.
java -jar $GATKjar -T VariantAnnotator -R $bundlePath/b37/human_g1k_v37_decoy.fasta -V recalibratedVariants.postCGP.Gfiltered.vcf -A PossibleDeNovo -ped trio.ped -o recalibratedVariants.postCGP.Gfiltered.deNovos.vcf
The annotation output will include a list of the children with possible de novo mutations, classified as either high or low confidence.
See section 3 of the method article for a complete description of annotation outputs and section 4 for an example of a call and the interpretation of the annotation values.