Hey 👋 I'm working on the

gridss/assemble

module. GRIDSS runs as a multi-step process (BWA index → preprocess → assemble), so the assemble step needs the output of the preprocess step as input. There's currently no suitable data for this in the nf-core/test-datasets repo, so to get the nf-test passing I'd like to upload a small .tar.gz of a staged preprocess directory to test-datasets (modules branch). Just checking in before I open a PR there - does that sound okay? Thanks! 🙏

Sounds perfect. The alternative is to use a setup block in the nf-test, where it runs gridss/preprocess.

Thanks Simon! Quick question: I'm also working on gridss/call, which needs the outputs of both preprocess and assemble. To test it, the setup block would chain three processes — BWA index → preprocess → assemble. Is that okay, or is chaining that many steps in a setup block frowned upon ?

it can be done, but does make it a bit of a pain

Thanks, Simon! I actually prepared the test data, but I realized I would need to run

untar

twice just to set up both the preprocess and assemble directories. Because of that, I thought chaining the processes in the setup block might end up being cleaner. The compute time for chaining is really fast anyway, so it doesn't add much delay. pls let me know which approach you think is best!

there are plenty of modules with a three process setup

Perfect, that makes things easier. Thanks for the quick response !!