Add msisensor (#95)

[davidmasp]

nf-core/sarek pull request

Description

Adding MSIsensor as a tool to Sarek. (from here #95 )

First nf-core commit so I am a bit clueless of what I am doing. Basically I added code to the pipeline to run msisensor which detects MSI status from NGS tumor/normal pairs. It's divided in 2 steps which I would say fit very nicely in Sarek. I guess this is a WIP PR. Will update further but thought it would be better to submit already.

PR checklist

• This comment contains a description of changes (with reason)
• If you've fixed a bug or added code that should be tested, add tests!
• Ensure the test suite passes (nextflow run . -profile test,docker).
• Make sure your code lints (nf-core lint .).
• Documentation in docs is updated
• CHANGELOG.md is updated
• README.md is updated

Learn more about contributing: CONTRIBUTING.md

[davidmasp]

I did build the environment in conda and I managed to run the nextflow run . -profile test --tools msisensor -resume and it does run but I don't have docker installed rn. I could try to configure it when I have more time though.

[davidmasp]

For the test I am not sure what to do, should I just include a test profile like conf/test_msi.config and there add tools msi? I mean, does it also check the output or just if the pipeline runs.

Regarding the output, the tool runs but it detects 0% which makes sense because the test dataset is small so I was also wondering if I need to build a test dataset to get a "supervised" output or just the fact that it runs is enough? I am not sure how I would add microsattellite inestability in a bam file, I guess it's possible though.

[maxulysse]

For the test I am not sure what to do, should I just include a test profile like conf/test_msi.config and there add tools msi? I mean, does it also check the output or just if the pipeline runs.

Regarding the output, the tool runs but it detects 0% which makes sense because the test dataset is small so I was also wondering if I need to build a test dataset to get a "supervised" output or just the fact that it runs is enough? I am not sure how I would add microsattellite inestability in a bam file, I guess it's possible though.

For now, let's just try adding MSIsensor to the list here:

sarek/.github/workflows/ci.yml

Line 134 in 52bee46

tool: [Haplotypecaller, Freebayes, Manta, mpileup, Strelka, TIDDIT]

We'll figure something out later.
We have an issue here to make some test data for Mutect2:
#148
I'm hoping that would be a solution for MSIsensor testing as well.

[maxulysse]

Thanks a lot for this PR, there's just a couple of things missing, but it's very promising.
I've been willing to add this tool for a long time.
I'm really happy that you did it

[davidmasp]

Okay, I think mostly I understand what i need to fill up

[davidmasp]

Will update when I have time, thanks!

[davidmasp]

For now, let's just try adding MSIsensor to the list here:

sarek/.github/workflows/ci.yml

Line 134 in 52bee46

tool: [Haplotypecaller, Freebayes, Manta, mpileup, Strelka, TIDDIT]

okay, done in 9afaa94

[davidmasp]

How should I update the readme?

[maxulysse]

How should I update the readme?

Add your name to the list of helpful contributors ;-)

[maxulysse]

But I guess you were speaking about the docs, and not the actual README file.

Basically, I would add a section in Variant Calling for microsatellites.
And then as we do for any other tool, just describe the actual tool (usually, I copy from their readme), and the output files.

[davidmasp]

Ah okok!

[davidmasp]

so only docs missing, I wanna check what I write is helpful and true so I'll take some time, not much I guess though.

[davidmasp]

everything should be more or less done I think just lmk otherwise. The test isn´t passing because it´s not using the new docker built right? I am still not completely sure how this works.

Cheers

[maxulysse]

That looks perfect, I just made a couple of more suggestions.
I think it should be MSIsensor according to their GitHub repo, and I prefer having MSIsensor as the output directory, as it's the name of the tool.

[maxulysse]

I'll push to dockerhub a new dev container containing msisensor and relaunch the tests as well

[davidmasp]

okay great! the MSI directory was to match the VariantCalling directory. As strelka is inside the variantcalling directory for instance. Does make sense? maybe the msisensor could go there too idk. Now is going directly to the results root directory?

[maxulysse]

Oh, yes, sorry, I don't have my whole head on the code today.
ASCAT results are in the VariantCalling directory as well, so I'll put MSIsensor results there as well for now

[davidmasp]

Awesome! now I think all is good no?

[maxulysse]

Awesome! now I think all is good no?

Yes!!!
Just waiting for the CI results and we're good 👍

[davidmasp]

nice!