Mutect2 does not need normal BAM anymore?

February 27, 2020 16:04
Edited

a) GATK version used: gatk v4.1.4.0
b) Exact GATK commands used

gatk --java-options "-Xmx32g" Mutect2 \
     -R $ref \
     -L $interval_list
     -I tumor.bam
     --germline-resource somatic-hg38-af-only-gnomad.hg38.vcf \
     -pon $pon \
     --f1r2-tar-gz $out/OB/${sample}.f1r2.tar.gz \
     --native-pair-hmm-threads 12 \
     -O $out/Unfiltered/${sample}.unfiltered.vcf.gz

The updated workflow here does not use normal BAM file while calling variants using Mutect2.

Do we not need normal BAMs for calling variants anymore as long as we have PoNs? I just wanted to confirm if that's the case.

Thanks!

5 comments

David Benjamin

February 27, 2020 17:23
Mutect2 has always supported a tumor-only mode, but it remains preferable to use a matched normal whenever possible.
1

Comment actions Permalink
Pan Canglong

March 13, 2020 17:19
Hi, David. I have two questions.

1.Do you mean that the tumor-only mode with germline mutation filtering is equal to tumor-normal paired mode?

2.the germline mutation filtering needs the germline.vcf, how can I choose from these which I find in the pagehttps://github.com/broadinstitute/gatk-docs/blob/master/blog-2012-to-2019/2018-01-31-Mutect2_resources_guide.md?id=11337

Thanks a lot!
0

Comment actions Permalink
David Benjamin

March 13, 2020 20:48
Pan Canglong

1. Tumor-only mode is far inferior the tumor-normal mode, regardless of germline filtering. It's the best we can do, but a matched normal is powerful in a way that a germline resource can't replace.

2. The pon is for the -pon argument, the af-only-gnomad is for the -germline-resource argument, and small_exac_common is for the CalculateContamination.
0

Comment actions Permalink
Pan Canglong

March 14, 2020 01:48
Thank you for your quick reply!

I have a new question to ask you. If I have another mutation results(.excel or .csv)from a standard nomral dataset but not paired to the tumor samples, can I use them to futher filter the mutation we acquired by the tumor-only mode step with germline filtering in GATK? Can I just eliminate the mutation both in our results from only tumor and the normal standard mutations?
0

Comment actions Permalink
David Benjamin

March 14, 2020 23:59
You could do this, but it would only amount to a weaker version of using gnomad for the germline resource. For that reason I would not recommend this approach.
0

Comment actions Permalink