FilterByOrientationBias vanished in 4.1.6.0
Hi,
I just tried out the new version 4.1.6.0 for our pipeline, however it seems that there is missing the 'FilterByOrientationBias' tool. So it breaks our pipeline:
A USER ERROR has occurred: 'FilterByOrientationBias' is not a valid command.
Is this on purpose? Can it be re-added?
Thanks
Dietmar
-
Hi riederd
FilterByOrientationBias has been removed from the latest release.
-
Is there any additional information about that? While I realize it was an "EXPERIMENTAL" tool, it would be nice to know why, and have some mention about that in the release notes.
Has the functionality been moved somewhere else?
-
A colleague noticed that there's now a ReadOrientationFilter in the Mutect Code . How does one use that?
I don't see that mentioned in the command line for FilterMutectCalls or Mutect2. How does one use it?
Perhaps with FilterMutectCalls --orientation-bias-artifact-priors,-ob-priors:File
One or more .tar.gz files containing tables of prior artifact probabilities for the read?
-
Hi,
Any arguments that are available can be found in the tools docs here: https://gatk.broadinstitute.org/hc/en-us/articles/360041414452--Tool-Documentation-Index
-
Bhanu Gandham, that doesn't answer the underlying question though: Has the functionality of FilterByOrientationBias been reimplemented somewhere else in the GATK? If so, how does one use it?
-
Sorry I hit enter before typing out the entire answer.
There is a new version of FilterByOrietationBias in v4.1.6.0. You can do the same by setting "run_ob_filter" to true in the mutect wdl https://github.com/broadinstitute/gatk/blob/master/scripts/mutect2_wdl/mutect2.wdl
-
Thanks Bhanu Gandham.
By digging into the WDL, it appears that the orientation bias filter is available at the GATK command line too. Here is line 597 in the WDL:
~{true='--f1r2-tar-gz f1r2.tar.gz' false='' run_ob_filter} \Which, I believe, means that if I add the following to my gatk Mutect2 call, I get the output file f1r2.tar.gz:
gatk Mutect2 --f1r2-tar-gz f1r2.tar.gz [...]
Subsequently, that gets passed into the gatk LearnReadOrientation call which generates "artifact-priors.tar.gz"
And that file gets into the FilterMutectCalls thusly
gatk FilterMutectCalls --ob-priors artifact-priors.tar.gz [...]
Is that all correct?
-
That is correct.
-
Bhanu Gandham, Am I correct that the OxoG and FFPE filtering has been removed from Mutect2/GATK?
I'm looking at the mutect2_wdl documentation, and it lists "Mutect2_Multi.artifact_modes" and "Mutect2.artifact_modes" as options to pass into their respective wdl's, but I can't find any reference to those parameters in any of the actual .wdl files there.
Previously, we'd been using the FilterByOrientationBias with the following options:
gatk FilterByOrientationBias \
--pre-adapter-detail-file sample.pre_adapter_detail_metrics.*\
--artifact-modes G/T \
--artifact-modes C/T [...]Is that no longer supported?
-
Take a look at this doc for info on the new filtering model for Mutect2, https://gatk.broadinstitute.org/hc/en-us/articles/360035531132--How-to-Call-somatic-mutations-using-GATK4-Mutect2
-
@riederd alanhoyle the old FilterByOrientationBias tool was based on a hokey non-probabilistic model. The new read orientation artifact model, which as Bhanu confirmed is part of FilterMutectCalls, is more sophisticated and more accurate. You no longer need to specify which artifact modes you wish to filter because the model automatically considers all possible substitution artifacts and filters appropriately. eg If your sample is FFPE, it will figure that out.
Please sign in to leave a comment.
11 comments