@Felix Krueger @Jonathan Manning moving here as #C05AHQPA9A7 thread is not really the right place 😆 https://nfcore.slack.com/archives/C05AHQPA9A7/p1777984771810759?thread_ts=1777982753.023139&cid=C05AHQPA9A7 Do we still need a separate FastQC task? Can TrimGalore do both pre- and post- trimming reports?

Right now the relevant subworkflows put UMI handling between the FASTQC and the trim step: https://github.com/nf-core/rnaseq/blob/master/subworkflows/nf-core/fastq_fastqc_umitools_trimgalore/main.nf

hmm, not sure about the pre-FastQC task? It probably can but I would have to instruct it to allow this

Hmm, not aware of anything, I've been running lots of tests, but I'll take a look.

yes we’ve got our own version of --genome +iGenomes

Ahh, that's clashing with some logic I recently added. Let me have a think

ERROR: unrecoginzed parameter name "genomeType" in input "genomeParameters.txt"

is indicative of an incompatibility of the STAR index version (i.e. the version of STAR used to build it) with the version of STAR used to run it. (It is one of my pet peeves with STAR who often that occurs, and how unnecessary this error is.) In this particular case you can just edit your

genomeParameters.txt

in the index and comment out (by prefixing

#

) the lines with

genomeType

,

genomeTransformType

, and

genomeTransformVCF

.

Is that really all you have to do? We have a whole load of logic in the #CE8SSJV3N pipeline that packages and runs with old versions of STAR if iGenomes is used with old versions of STAR. Maybe we should just have a

sed

command before running! 😅

It depends on the versions (as mentioned, this happens once in while). There are some old index version that were built with a totally principle/algorithm and are completely incompatible. But the change above is I believe newer and could have been avoided by better design in STAR (just ignore lines you don't understand or that are now obsolete; e.g. Nextflow is fairly good at doing that), and I have made an index compatible by literally just commenting these lines in this one file. If you want to build a

sed

command, you would probably have to check the version of the index and compare that to the STAR version to be sure that this is possible in this particular case ...

Yep, the way Felix uses the workflow with a custom genome config wasn't a pattern I'd properly accounted for, meaning that he was supplying a modern index but the pipeline was sending reads to the old STAR. Fixed it earlier on: https://github.com/nf-core/rnaseq/pull/1834

I am very fuzzy on this, because I just build STAR indices every few years when there is a new mouse or human genome assembly, or a new version of STAR that necessitates this, but I think those lines are optional and new versions of STAR generate them but are fine without them, whereas slightly older versions choke on them. But as I said, there are some incompatible changes if you go further back, so it all depends on the exact versions.

I'll look into it. I wouldn't be bothered to wade into it myself, but claude can probably answer the question

shared a GIF

Cool! This was pure coincidence -- when I saw the error message, I thought "STAR version incompatibility" and went to a random index I had made to verify. Looking at that I noticed I had made a backup version of the

genomeParameters.txt

and compared that to the current version, noticing the commented-out lines. Thank your for digging into this and taking an off-hand statement to the next level! 🤝

Claude did all the work while I had a cheeky whisky