Bump up kb to 0.28.0

[gennadyFauna]

Description of feature

The new version of kallisto|bustools is out (https://www.biorxiv.org/content/10.1101/2023.11.21.568164v1): kb is at 0.28.0, breaking compatibility with the old versions. The nucleus and lamanno workflows have been superseded by nac, which separately quantifies nascent, mature, and ambiguous transcripts. In addition, kb count now accepts an optional sum argument, which adds matrices together to match common workflows.

It would be good to move nf-core/scrnaseq to the newest version. I would appreciate help; the right way to combine the sum variable and the h5ad generation is not obvious to me.

[grst]

Hi @gennadyFauna,

thanks for the heads-up, I'm all for updating this!

I would appreciate help; the right way to combine the sum variable and the h5ad generation is not obvious to me.

I suggest to first bump the version of kb in nf-core/modules. Then make a PR to scrnaseq updating the module and we'll see what exactly breaks.

[Yenaled]

@gennadyFauna -- I am an author on the preprint. Latest version of kb is 0.28.2; deff update to that if you can!

For standard workflow, commands are the same as before.

For the nac index type, --sum=total should be used and the resulting cells_x_genes.total.mtx matrix will give you the CellRanger7-like quantification (i.e. introns are counted) whereas the cells_x_genes.cell.mtx matrix will give you the kb standard workflow quantifications which are essentially the CellRanger3-like quantification (i.e. introns not counted).

For single-cell RNA velocity type analysis, use cells_x_genes.cell.mtx for your "spliced" matrix and cells_x_genes.nascent.mtx for your "unspliced" matrix.

Side note: You can also use kb-python for bulk RNAseq (so might be worth updating the bulk RNAseq pipelines for that as well ;) -- the abundance.tsv output will be the same as before and can be thrown into DE programs).

[grst]

#294 has been merged in dev and will be part of the next release.