If not an error, choose a category for your question(REQUIRED):
c) Why do I see (......)?
GATK version: 18.104.22.168
I'm running Mutect2 over samples for which I have provided a VCF of alleles that I would like force-called. In 99% of cases this works great, but in a few instances a couple of alleles that I have included for force-calling do not actually make it into the output calls VCF.
My command is:
gatk --java-options -Xmx3000m Mutect2 \
-R /data_in/MY1776/MY1776.self.ref.shifted_by_8000_bases.fasta \
-I /data_in/MY1776/self_realigned_shifted.bam \
--read-filter MateOnSameContigOrNoMappedMateReadFilter \
--read-filter MateUnmappedAndUnmappedReadFilter \
-O /out_default.vcf \
--alleles /data_in/MY1776/MY1776.self.ref.reversed.selfRef.shifted.homoplasmies.vcf.bgz \
--annotation StrandBiasBySample \
--max-reads-per-alignment-start 75 \
--max-mnp-distance 0 \
-L chrM:8023-9140 \
--debug-assembly-variants-out /rej.vcf \
In this instance the variant in question is listed in the rej.vcf file obtained via `--debug-assembly-variants-out`. I have examined `bamout.bam` as well as the input bam and there appears to be ample coverage at the site of interest (the T at position 8316 is the position of interest, highlighted):
I have tried running this with some of the additional parameters in https://gatk.broadinstitute.org/hc/en-us/articles/360043491652-When-HaplotypeCaller-and-Mutect2-do-not-call-an-expected-variant (namely `--linked-de-bruijn-graph` and `--recover-all-dangling-branches`) to no avail. Coverage is very deep at this position (>2000x). Notably if I edit the input to `--alleles` and change the allele of interest (8316:T>A) to anything else (8316:T>C or T>G) it appropriately shows up in the output VCF.
What am I missing here? Let me know if you have any solutions or if you need any additional files.
UPDATE: Adding `--disable-adaptive-pruning` now produces the variant of interest specified in --alleles, but also adds several other new calls, in case that is helpful in isolating where this force-call variant is being lost.
Please sign in to leave a comment.