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.

Triallelic sites in MuTect

3141531415 n/aMember


I am running MuTect 1.1.7 on some amplicon data that targeted several specific genes. There's one documented mutation in particular that MuTect correctly identifies, but rejects due to, "triallelic_site," according to the extended output. I have two questions about this rejection:

  1. The description of filters from the 2013 Nature paper says this flag is raised when the normal sample is heterozygous and mutect is considering another alt allele. How am I seeing "triallelic_site" if I am running MuTect without a normal?

  2. Is there a way to stop mutations from being rejected due to triallelic site? I tried passing mutect the --force_alleles option and it successfully marks my particular mutation of interest as KEEP in the mutect.out. However, using --force_alleles causes empty vcf files that only contain the header info. Is --force_alleles supposed to work with MuTect's vcf output?


java -Xmx4g -jar mutect-1.1.7.jar -T MuTect --reference_sequence genome.fa --enable_extended_output --cosmic cosmic.vcf --dbsnp dbsnp.vcf --input_file:tumor input.realign.recal.bam --vcf mutect.vcf --out mutect.out --coverage_file coverage.wig --gap_events_threshold 50 --fraction_contamination 0 --pir_median_threshold 0 --force_alleles



  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Hi Karl,

    I'll have to check with the developer, but I think MuTect may be considering the alleles in the dbsnp file (which is used as a sort of proxy for the normal when you don't have one).

    There's no straightforward way to disable the default filters. I suspect the empty vcf generated when you use --force_alleles is a bug.

Sign In or Register to comment.