just a rustqc versions bump: https://github.com/nf-core/modules/pull/11143

already?

Blame @Phil Ewels 😄

Poor him, I argue with him enough already

Good point, well made

Not yet 😆

@Jonathan Manning Am I right into assuming that I can use RUSTQC to replace samtools stats, flagstat and idxstats, so basically the whole BAM_STATS_SAMTOOLS subworkflow?

Bot response:

Yes, that's correct - RustQC produces all three outputs (`flagstat`, `idxstats`, and `stats`), so it can replace the whole `BAM_STATS_SAMTOOLS` subworkflow.

- **flagstat** and **idxstats** are numerically identical to samtools
- **stats** matches on all Summary Numbers (SN) fields and 19 of 20 histogram sections - the only minor difference is in the GCD (GC-depth) section, which can show tiny floating-point variation (~0.018 max) due to sampling, but nothing meaningful in practice
- All outputs are MultiQC-compatible, so downstream reporting works as-is

These are all generated as part of RustQC's single-pass BAM processing, so there's no additional runtime cost - you get them alongside featureCounts, dupRadar, RSeQC, and Qualimap outputs in the same run.

OK, so we need to patch all subworkflows containing this BAM_STATS_SAMTOOLS subworkflow

hold off until we've trialled it a bit

Any other way to trial than pushing in prod?

@Maxime Garcia check that it actually improves things first - Samtools is pretty efficient already

We are running 3 times samtools to get that

You can turn stuff off in RustQC - it produces all outputs by default but you can give it a config file / env vars / CLI flags to turn bits off that you don't want

yeah, sorry, need to read the docs

Which is why you can also do it with env vars and CLI flags 😉

Eagerly awaiting the

rustqc dna

command 😉

For samtools (flagstat, stats, idxstats), I think it's one of the first tool we considered to group together as a single module

Yes, viewing subworkflows as potential targets is a pretty good idea!

I actually factored out the relevant rnaseq steps into a subworkflow so that the switch with RustQC was easier to read, I should probably make that an official nf-core subworkflow

That would facilitate my plan to integrate it in #C01NHS4CC91

Here you go! https://github.com/nf-core/modules/pull/11147

yeah I noticed

just noticed the versions channel needs stripping

oh

I just spotted alignments you won't like, will fix

I just let LS do it for me, and I gave up on upholding Harshil's alignment

Not a bad plan, I'm in favour of stripping conditionals for channel logic. Coming up...

I did that in #C06PHJM0Y3D -> https://github.com/nf-core/seqinspector/blob/dev/workflows/seqinspector.nf I only have a single tools array, that contains tools to be run, so logic is much clearer

OK done, tests running

It does look SO MUCH nicer

Thanks! Will do the workflow install soon

Only last thing I can see remove params from the subworkflows