A community effort to collect a curated set of analysis pipelines built using Nextflow.

nf-core

Hi  :wave:
Working on the `snpclustering` subworkflow (PR #11879 — <https://github.com/nf-core/modules/pull/11879|github.com/nf-core/modules/pull/11879>) and running into unstable md5 sums in the nf-test snapshots. Some outputs (UMAP/t-SNE PNG plots, BEAGLE5/PLINK2 log files) seem to vary between runs, likely due to embedded timestamps or non-deterministic rendering.

How do you usually handle this: exclude these outputs from the snapshot, or just check existence/filename instead of content? Can someone help me out with this? Thanks!

The `sanitizeOutput`  function from nft-utils allows you to say that certain channels are unstable: <https://nf-co.re/nft-utils#sanitizeoutput---sanitize-process-output-to-create-clean-snapshots|nf-co.re/nft-utils#…>

so `snapshot(sanitizeOutput(workflow.out, unstableKeys: ["foo", "bar"])).match()`

Thanks <@U016SFULDB8>, for `tsne_tsv`/`umap_tsv` specifically though, would `nft-csv` be a better fit than just marking them unstable? They're numeric coordinate outputs, so ideally we'd compare with some floating-point tolerance rather than excluding them from comparison entirely