Target variants not in active region
Answered
We have a known SNV with AF ~1% and are trying to use Mutect2 to detect it in amplicon sequencing data.
- It can be called with no problem in some samples but cannot be called in other samples.
- The reads support ~1% AF for the SNV when examining input bam files.
- When running in debug mode, the log revealed that the region is not considered active and the variant was not called by Mutect2.
- The variant can be called with no problem with --force-active on.
We wonder how we can call the expected SNV in all samples consistently. One way we are thinking is to adjust active threshold. We also tried to examine activity profile but somehow --activity-profile-out is not available in GATK 4.1.8.1.
-
Hi yqiu,
I am going to move your post into our Community Discussions -> General Discussion topic, as the Somatic topic is for reporting bugs and issues with GATK.
You can read more about our forum guidelines and the topics here: Forum Guidelines.
Best,
Genevieve
-
Hi yqiu,
It's generally pretty advanced to change the options that affect the active region determination, but there are a couple options you can try:
- --initial-tumor-lod
- --active-probability-threshold
You can read more about these in the Mutect2 tool docs page: https://gatk.broadinstitute.org/hc/en-us/articles/4418062771227-Mutect2.
The option --activity-profile-out is no longer available in Mutect2, we have some newer debugging options. Here is an article describing these troubleshooting options and steps: https://gatk.broadinstitute.org/hc/en-us/articles/360043491652-When-HaplotypeCaller-and-Mutect2-do-not-call-an-expected-variant
Please let me know if you have further questions.
Best,
Genevieve
-
Hi Genevieve,
Thanks for the detailed response!
Do you have any suggestions about which parameter to adjust and suggested values? Or should I adjust both of them?
For --initial-tumor-lod, should I increase it or decrease it? We don't have normal samples. If so, is TLOD still relevant?
For --active-probability-threshold, I am not sure which value will be proper. That's why I want to --activity-profile-out to see the region's activity score. If you have any recommendations, it would be very helpful.
-
Yes, you can still use --initial-tumor-lod even though you don't have normal samples. I think it would be best to adjust these parameters one at a time at first and see how that affects your results and this variant of interest. Decreasing the initial tumor lod should be able to decrease the threshold for variants to be considered active.
The --active-probability-threshold is already quite low for Mutect2 (0.002). You can try decreasing it a bit, but potentially you'll just want to try keeping --force-active on when you are trying to look for SNPs with such a low AF.
-
Thanks for the quick response and suggestions!
I tried them separately. Unfortunately, neither parameter can rescue the variants. I tried --initial-tumor-lod 1 and --active-probability-threshold down to 0.0001 respectively. So far it looks like the only way to rescue it is to use --force-active.
Could you help explain more about the relationship between --initial-tumor-lod and --active-probability-threshold? I am not sure how they affect each other.
And do you have any further suggestions except for --force-active? Should I try lower values or other parameters? I don't think 1% AF is a very low target though I might be wrong.
Thanks again for all the helpful advices!
-
Hi yqiu,
Apologies for the delay, I was out of the office last week!
I followed up with our developers to get more specific details regarding --initial-tumor-lod and --active-probability-threshold. I found out that --active-probability-threshold does not make a major difference with the active region determination, --initial-tumor-lod will help the most.
You can try --initial-tumor-lod -1 to see if that helps. If it does not help, there might be a strange problem coming up from your sequencing data. What are the base qualities of the reads in the region of this SNP? Is the sequencing high quality?
Let me know,
Genevieve
Please sign in to leave a comment.
6 comments