Create a sentieon specific subworkflow for DSL2

[maxulysse]

EDIT (from @cjfields): Modules checklist:

• [x] alignment (accelerated bwa)
• [ ] ~BQSR (optional?) (EDIT: Won't be implemented. It should be possible to use GATK-based BQSR.)~
• [x] DNAseq (SNV/Indel) - including GVCF support
• [ ] DNAscope (SV and SNV/Indel) - including GVCF support
• [ ] TNseq (tumor/normal) - including GVCF support
• [ ] TNscope (tumor/normal)

[cjfields]

Checklist (some are already in place)

Modules

• [x] alignment (accelerated bwa)
• [ ] BQSR (optional?)
• [x] DNAseq (SNV/Indel) - including GVCF support
• [x] DNAscope (SV and SNV/Indel)
• [ ] TNseq (tumor/normal) - including GVCF support
• [ ] TNscope (tumor/normal)

[FriederikeHanssen]

BQSR would be done also with a Sentieon subtool or GATK?

[maxulysse]

Sentieon has that if I remember well

[cjfields]

Sentieon has that if I remember well

Correct; it also supports VQSR though I'm not sure if that's on the roadmap here, I've been running it separately.

[FriederikeHanssen]

carefully toying around with it, is the correct answer I would say.

[nicorap]

we have a few modules ready from the danish national genome center, I'll just a add them and make the missing ones also with a sub workflow.

[cjfields]

Small note: in the 2.7.2 workflow we found a bug which skips Sentieon DNAscope and DNAseq when using manually recalibrated BAMs, which seems to be due to a difference in how GATK and Sentieon steps are implemented. GATK HaplotypeCaller, Manta, etc) appear to use a recalibrated BAM file with index here.

However, the next line of code indicates Sentieon steps are using the deduped BAM + index + recalibration table, and the related processes require that table. So, when given a recalibrated BAM + index, this effectively skips Sentieon calling altogether. The input channel for those steps also isn't generated when given a table like this.

The reason I point this out: we can add a simple fix for this and test it, but it's worth noting that all others steps apart from Sentieon use an already recalibrated BAM file and the Sentieon BQSR generates this BAM file anyway, so I'm not sure of the benefit of having this oddly unique step just for Sentieon. Maybe something to consider for the DSL2 implementation?

[nicorap]

That makes sense, and also explain why I never got to have sentieon running using 2.7.2 from bam. 😐. I think it’s important to be able to skip BQSR, as for production setups the data should always be the same, so it’s not needed.

[maxulysse]

@asp8200 I feel we're getting close to finalize this issue

[asp8200]

TNseq and TNscope still not implemented :-/

[maxulysse]

yeah I know, but we're getting closer and closer to completion

[cjfields]

@maxulysse you are awesome for taking this one, we're about to get this running for some of our projects.

EDIT: just to add, we're not using TNseq/TNscope for this one (it's all germline) but we can certainly try it out.

[maxulysse]

@asp8200 did all the work, he should get the praise. You should be able to test it, it's in the latest release already