Open migbro opened 3 years ago
@migbro Are you just running Mutect2
, or are you running a pipeline consisting of multiple tools? If the latter, which specific tools are you running?
Also, you are running a very old release of GATK. Some of these old releases are known to have some rare sources of non-determinism in Mutect2
which could cause results to vary slightly across runs. We recommend upgrading to the latest release (4.2.0.0
) if you can, and using the official GATK docker images in dockerhub (https://hub.docker.com/r/broadinstitute/gatk/).
Hi @droazen , thanks for your quick response! Depending on what you mean for "multiple tools," I'd say just Mutect2. I say this because we are running this as part of a cwl workflow, so Mutect2 is running on its own instance (virtual machine) with that specified docker image. All of the surrounding tools have produced the expected outputs when queried except for Mutect2. That's a good point about the release version, I think we happen to be considering upgrading, but have to be picky as to when given the size and scope of our operation :) Also, your suggestion of using the official GATK might help, as I am not sure I can rule out subtle apt update
changes even within the same version of ubuntu over time.
Bug Report
Affected tool(s) or class(es)
Tool/class name(s), special parameters? Mutect2
Affected version(s)
Description
Describe the problem below. Provide screenshots , stacktrace , logs where appropriate. We have run mutect2 on the same sample using the same input crams, references, and intervals. The only discernible difference is the docker image used that we built. The difference between the first and second one is that the first one was built without samtools, the second one with samtools. The third is exactly the same as the second except it was re-built about a year or more later. Looking at a count of the
PASS
results based on each:Out of curiosity, we tried building the docker again without samtools, so in theory, the only possible change is that when each docker is built, apt update is run. The differences are small, but is that expected? That with and without samtools, and if
apt
packages change, mutect2 could be influenced?Steps to reproduce
Tell us how to reproduce this issue. If possible, include command lines that reproduce the problem. (The support team may follow up to ask you to upload data to reproduce the issue.) I can't replicate scenario 1 in the table because it was built in 2019, so apt packages were different then. Scenario 2 dockerPull:
docker pull kfdrc/gatk:4.1.1.0
Scenario 3 dockerPull:docker pull pgc-images.sbgenomics.com/d3b-bixu/gatk:4.1.1.0
Scenario 4 dockerPull:docker pull migbro/gatk:4.1.1.0L
No samtools Dockerfile:
Yes samtools Dockerfile:
Expected behavior
Tell us what should happen All
PASS
var counts are the sameActual behavior
Tell us what happens instead
PASS
var counts vary slightly +/- samtools and year docker builtThank you for your time!