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.

Variant recalibration tranche plot gives a very high number of novel variants and almost no FPs

Hi,

My question concerns the VQSR step. I am using GATK version 3.7 on 350 human WES samples. After the calling with HaplotypeCaller I have used the VariantRecalibrator function with the following command (see below) I get a tranche plot with a very high number of novel variants and almost none FPs compared to what I have seen on the best practice website. Ti/Tv values are ~2. Could you let me know if this is correct? Thanks!

Eric

java -jar $GATK_tools/GenomeAnalysisTK.jar \
-T VariantRecalibrator --disable_auto_index_creation_and_locking_when_reading_rods \
-R ${human_genome}_no_chr.fa \
-input $input/pool.annotated.sites_only.unfiltered.vcf \
--num_threads 25 -recalFile $output/pool.snps.recal_260918 \
-tranchesFile $output/snps.tranches_260918 \
-an QD -an MQRankSum -an ReadPosRankSum -an FS -an MQ -an InbreedingCoeff \
-resource:hapmap,known=false,training=true,truth=true,prior=15 $resources/hapmap_3.3.b37.vcf \
-resource:omni,known=false,training=true,truth=true,prior=12 $resources/1000G_omni2.5.b37.vcf \
-resource:1000G,known=false,training=true,truth=false,prior=10 $resources/1000G_phase1.snps.high_confidence.b37.vcf \
-resource:dbsnp137,known=false,training=false,truth=false,prior=2 $resources/dbsnp_138.b37.vcf \
-mode SNP \
-rscriptFile $output/snps.recalibration_plots_260918.rscript

Best Answers

  • etrepoetrepo
    edited October 2018 Accepted Answer

    Hi @Geraldine_VdAuwera ,

    Thank you very much for your help. Indeed, after using a --target_titv of 3.2, the FP proportion increases significantly. May I ask you, if the number of novel variants displayed on the tranche plot looks odd to you? It is much higher than what I have seen on the best practice website for similar plots.

    Thanks!

    Eric

    Post edited by bhanuGandham on

Answers

  • etrepoetrepo Member
    edited October 2018 Accepted Answer

    Hi @Geraldine_VdAuwera ,

    Thank you very much for your help. Indeed, after using a --target_titv of 3.2, the FP proportion increases significantly. May I ask you, if the number of novel variants displayed on the tranche plot looks odd to you? It is much higher than what I have seen on the best practice website for similar plots.

    Thanks!

    Eric

    Post edited by bhanuGandham on
Sign In or Register to comment.