added cellranger vdj/mkvdjref modules, with module test for vdj

[klkeys]

PR checklist

• This comment contains a description of changes (with reason).
• If you've fixed a bug or added code that should be tested, add tests!
• If you've added a new tool - have you followed the module conventions in the contribution docs
• If necessary, include test data in your PR.
• Remove all TODO statements.
• Emit the versions.yml file.
• Follow the naming conventions.
• Follow the parameters requirements.
• Follow the input/output options guidelines.
• Add a resource label
  - [ ] Use BioConda and BioContainers if possible to fulfil software requirements. only works with nfcore/cellranger Docker image

adds new cellranger module cellranger vdj for single cell immune profiling

mostly cribbed from cellranger count and cellranger mkref

includes the corresponding cellranger mkvdj ref just in case

test data included in nf-core/test-datasets#787

[SPPearce]

Is there a possibility to have a test for mkvdjref?

[klkeys]

Is there a possibility to have a test for mkvdjref?

@SPPearce regarding a test for mkvdjref, I'm a little torn

cellranger mkvdjref has two possibilities:

1. creating a ref from a GTF + a FASTA
2. using a custom fetch-imgt script supplied with cellranger

if I write a mkvdjref test, then I'm inclined to support the 1st but not the second option, since the fetch-imgt script is controlled by 10x

does that sound ok? if so, then I'll draft a mkvdref test 👍🏼

[grst]

1. Agree on using the cellranger mkvdjref with fasta/gtf. The fetch-imgt script requires an internet connection and in general our workflows should run offline. People can still run it independently and specify it as reference

2. Is there still any reason to run cellranger vdj over cellranger multi? For paired vdj+gex, cellranger multi is the recommended way. If there's only VDJ, cellranger multi can also be used. That way we would save one additional alignment route in the scrnaseq pipeline:
   
   (https://support.10xgenomics.com/single-cell-vdj/software/pipelines/latest/using/multi)

[klkeys]

@grst no reason to prefer cellranger vdj over cellranger multi, I started with V(D)J simply because it met a need

but I'll start with cellranger multi once this is done

in case it matters, cellranger multi is different from count and vdj in that it uses a strange CSV-like input format

it will be some work (or maybe unwise to even try) to shoehorn into current nf-core samplesheet standards 😕

flipside is that a cellranger multi module wraps a fundamentally simple command:

    cellranger multi --id ${meta.id} --csv ${meta.cellranger_multi_csv_file}

and then cellranger handles all the other complexities (gene expression, feature barcode, cell multiplexing, V(D)J profiling, etc.)

[grst]

in case it matters, cellranger multi is different from count and vdj in that it uses a strange CSV-like input format it will be some work (or maybe unwise to even try) to shoehorn into current nf-core samplesheet standards confused

I don't see a reason why we can't generate these files on a per-sample basis. In fact, @ErikaKvalem has started to play around with this. It would, of course, require an extension of our samplesheet, but there was some discussion (and a proposed solution) here: nf-core/scrnaseq#174 (comment)

[klkeys]

added requested test for cellranger mkvdjref 👍🏼

[SPPearce]

Please don't merge just yet, I'm just looking at something here.

[SPPearce]

I have updated the modules to use the prefix variable instead of meta.id for the output. This is set to be meta.id by default, but lets you change the filenames via a config.
I have also removed the --sample $meta.id argument, which the tools are using to specify which fasta when there are multiple inside the folder provided, rather than using all of them. I think this is was likely to cause a failure anytime that the meta.id was set to be anything but the prefix of the fastq filename.
Tests were passing on gitpod with docker, so hopefully still passing here.

[SPPearce]

I've made changes with respect to the prefix, hopefully will all pass and then I'm happy.

[adamrtalbot]

Could you use the full URI for the docker images please?

[SPPearce]

@adamrtalbot , I have just updated to use the full path in all the modules.

[SPPearce]

Right, I'm happy to go now. @adamrtalbot , if that was your only objection I think we are happy to merge?

[adamrtalbot]

apologies for being slow!

[klkeys]

oof this sparked a lot of discussion 😵

thank you for the additional eyes and edits @SPPearce @jfy133 @adamrtalbot

I will proceed with merging now 👍🏼

[SPPearce]

Thanks for your hard work and patience @klkeys

[klkeys]

premature celebration? it looks like nf-core linting failed on cellranger/mkref and cellranger/mkvdjref 😣

not sure how this escaped me, but best not dwell on the past, I'll correct this in #3229

[SPPearce]

That's strange, why did the linting tests not run? All the tests that ran were passing.

[klkeys]

no clue 😣

see here: https://github.com/nf-core/modules/actions/runs/4832726781/jobs/8611869687

I've fixed it in #3229