Re: Worflow langages

[Arun Isaac]

Hi all,

>> - CWL implementation (like CCWL recently)
>> - ravanan
>
> Arun is doing cool stuff in this area.  IIUC, there is: 1. CCWL [1]
> where one describes the workflow using a Scheme DSL that is then
> “compiled” to CWL language thus processed by some CWL engines.
>
> And 2. ravanan [2] where one describes the workflow using the CWL
> language.  In other words, ravanan is yet another CWL engine, but this
> one backed by Guix.
>
> Somehow, ravanan brings the ideas behind GWL – reproducible workflow
> require to track all the details of the deployment – and relies on CWL
> specification instead of yet another Workflow Language. ;-)
>
> 1: 
> https://hpc.guix.info/blog/2022/01/ccwl-for-concise-and-painless-cwl-workflows/
> 2: https://github.com/arunisaac/ravanan

Yes, that's correct. ccwl is a compiler that generates CWL. And, ravanan
is a "runner" that executes CWL. Both projects can be used completely
independent of each other. You can use ccwl to generate CWL for any CWL
standards-compliant runner (not just ravanan). And, you can use ravanan
to run any CWL workflow---hand-written or generated.

>> On the last Guix-HPC report, UTHSC  seems to experiment with CCWL for
>> pangenomic data. Are there some feedback on that yet ?

I developed ccwl when I was working for UTHSC. I have since moved on to
University College London. I continue to use ccwl for my own
bioinformatics pipelines (specifically in pangenomics). Pjotr Prins from
UTHSC (mentioned in the Guix-HPC report) is a long supporter of the
project, and has expressed interest in using it. I also presented ccwl
at MemPanG24, and there was some interest. But, other than that, I am
not aware of any large community of users.

Regards,
Arun