Adam Talbot

We made it! 🥳

Thanks Rob, that's really helpful. I think we should allow users to set it based on how they want, ideally with the default remaining backwards compatible. We're not exactly sure...

@maxulysse is this still valid?

Looking at this PR, we would be moving the logic of the publishing from the configuration to the pipeline. I'm quite happy with that, but: - I know some people...

OK I've read [this comment](https://github.com/nextflow-io/nextflow/issues/4670#issuecomment-1956040723) which has made it more clear. We add a second structure to Nextflow, the `output` block, which includes the definitions about what should be published...

This is beginning to look great. All the publishing logic is in one location, easy to review and understand where it's coming from. There are two downsides to this approach:...

> For example, QUANTIFY_PSEUDO_ALIGNMENT is used twice in RNASEQ, once as itself and once as the alias QUANTIFY_STAR_SALMON. One publishes to the folder "${params.aligner}" while the other publishes to "${params.pseudo_aligner}"....

This follows on from #1194 and #1184. Once they are complete we should see if all these tests are still required.

Summary of existing tests (by eye): ### Skip steps --skip_qc --skip_trimming ### Run without reference genomes --gtf false --star_index false --transcript_fasta false --rsem_index false (STAR-RSEM) --rsem_index false (HISAT2) ### Alternative/additional...

# Proposal ## Remove or replace with subworkflow level tests The following can be made into subworkflow level tests or removed entirely. - [ ] `PREPARE_GENOMES` https://github.com/nf-core/rnaseq/issues/1256 - [ ]...