Gatk haplotyper

Question

[prs]

If I am using Gatk haplotyper for for tumor and normal bam files, and then filtering the common ones to identify germline variants, is it the correct approach? And what difference will it make difference if I run haplotyper on tumor-only bam? Please suggest.

[Pierre Lindenbaum]

Don't forget to follow up on your threads. If an answer was helpful, you should upvote it; if the answer resolved your question, you should mark it as accepted. You can accept more than one answer if they all work. If an answer was not really helpful or did not work, provide detailed feedback so others know not to use that answer.

• Python for pdf reports
• Denovo transcriptome assembly
• RPKM normalization
• Biological replicates
• Strand non-specific bam file
• Bam to bedgraph

[prs]

I checked and accepted for all threads. Thank you.

[QX]

I think they will compare the the reference if you don't provide the normal samples

Answer 1

[Michael]

Mutect2 is the tool from the GATK toolbox that is designed for detection of somatic variants. And this is the best-practice work-flow for germline SNVs based on HaplotypeCaller. See also: Somatic calling is NOT simply a difference between two callsets

In summary:

Calling somatic variants: Mutect2 with tumor.bam + normal.bam (if you have it)
Calling germline: HaplotypeCaller with normal.bam only

[prs]

Thank you, I understand the difference between the two approaches now. My question pertains to the initial phase, specifically regarding the use of haplotyper. I’d like to clarify a few points for both scenarios: Tumor-normal BAM: If I run haplotyper on both tumor and normal BAM files simultaneously, is it appropriate to filter out variants that are common to both samples ( to get only germline)? Tumor-only BAM: Will using haplotyper on a tumor-only BAM file yield any significant differences compared to the tumor-normal approach?

Since I’m new to this process, I wanted to ensure I’m approaching it correctly before moving forward. Any guidance would be appreciated!

[Michael]

Tumor-normal BAM: If I run haplotyper on both tumor and normal BAM files simultaneously, is it appropriate to filter out variants that are common to both samples ( to get only germline)?

As I understand it, no it won't be correct. If you want germline use HaplotypeCaller on the healthy tissue, if you want somatic use Mutect2.

The two approaches have some important differences regarding allele frequencies and ploidy. Also, germline variants and stats can be called for a population of multiple samples, while somatic are from the same person. Germline callers will assume constant ploidy, while due to different mixtures this cannot be assumed for cancer samples. Somatic caller will also suppress reporting germ line variants where possible.

[prs]

I understand now. But, I have read it many times that tumor-normal is gold standard for calling germline variants. So how can we make the use of tumor bam in this approach?

[Michael]

Are you sure your source didn't say gold standard for "somatic"? The terms may be a little confusing at times. somatic ('body') = mixture of tumor + "normal cells", germline = inherited variants + de novo germline variants (ofc these are also sequenced from body cells e.g. blood).

[prs]

Actually these references mentioned - https://www.annalsofoncology.org/article/S0923-7534(22)04771-8/fulltext : "the merits of paired tumour-normal sequencing to ensure complete detection of all germline variants"; https://www.sciencedirect.com/science/article/pii/S0923753422000151 : "clinical germline testing for cancer patients still remains the ‘gold standard’ for the detection of these variants" But I am just not sure on how to use the tumor-normal in pipeline.

[Michael]

I think this text from the introduction doesn't say what you think. Look closely, highlight and word in parenthesis are mine:

Many advocate the merits of paired tumour-normal sequencing to ensure complete detection of all germline variants, avoiding allelic loss and including exon-level deletions/duplications and intronic variants.3,4 Nevertheless, for reasons of cost and logistic simplicity, (over) tumour-only sequencing remains the most frequent approach.5

So, this is a study assessing whether tumor-normal is better for uncovering important germline variants than tumor only. And this is a high-level study on sequencing in precision medicine, not the variant calling step in general.

[prs]

Yes, so from the perspective of precision medicine, how can we use paired tumor-normal in gatk?

[Michael]

Please read my answer again... Especially the first word :)

[prs]

But Mutect2 is for somatic variant calling. I only have a one question, that is how can I make use of tumor bam in germline variant calling? Or should I only consider normal bam?

[Michael]

What would you need the tumor sample to call germline variants if you have a normal sample. Use haplotypecaller for germline using the normal sample. The tumor sample won't add any more information about the germline to that.

Somatic: Mutect2 with tumor + normal.bam
Germline: HaplotypeCaller with normal.bam

[prs]

Ok. I was going with this approach only, but I thought it would be a kind of checkpoint to see variants in tumor bam also, as a germline variant should be present in tumor as well as normal bam. Thank you for clarifying.

[Michael]

If you think of the tumor as being under an infinite sites model, then yes, all germline variants will be present in the tumor. However, it isn't at all like that in reality, as larger variants and aberrations can be present in the tumor such that some regions with germline variants are deleted or have different copy-numbers. Also, you don't know the mixture proportions of tumor- and other cells in the samples , such that tumor alleles can have massively different coverage. But knowing true germline variants can help determine e.g. confidence intervals for genotypes.