new subworkflow: distributed computing GRIDSS subworkflow

[johnoooh]

Is there an existing subworkflow for this?

• I have searched for the existing subworkflow

Is there an open PR for this?

• I have searched for existing PRs

Is there an open issue for this?

• I have searched for existing issues

Are you going to work on this?

• If I'm planning to work on this subworkflow, I added myself to the Assignees to facilitate tracking who is working on the subworkflow

[johnoooh]

So there is currently a GRIDSS module in nf-core/modules but this method can be slow when running very complex tumor-normal pairs. In a HPC environment distributed computing can speed up this calling. GRIDSS has commands that allow for distributed computing but they should be broken up into different modules. Unfortunately these distributed jobs should be run in the same working directory, which is something nextflow cant really do due to the nature of the work directory. Many intermediate files are created by the individual steps and they must be passed from one process to another and then renamed to the proper names. I have done it this way and it works but I am of course open to suggestions on better ways to do this.

[famosab]

@johnoooh Do you still plan to work on this? If you need input hopping on over to slack might also be worth a try! :)

[johnoooh]

Oncoanalyzer now does these steps and integrated them into a subworkflow, but the modules are not in the nf-core modules repo. They are local modules for the pipeline. They need tests at least to be integrated into the nf-core repo. I'll look into doing this for the hackathon. Is this something the oncoanalyzer team would like? @scwatts what do you think? I know having it as a local module rather than a nf-core module gives you a bit more flexibility.

https://github.com/nf-core/oncoanalyser/blob/master/subworkflows/local/gridss_svprep_calling/main.nf

[scwatts]

Hi @johnoooh, we're no longer using GRIDSS in oncoanalyser after replacing it with ESVEE - let me know if there is still interest and I can open a PR in nf-core/modules that includes the GRIDSS processes I wrote to kick off this work!