Flexible Input Formats In our last meeting, Lukas and I discussed supporting different input formats for the count matrices. I researched how other pipelines handle this and checked the input requirements for our modules (see this issue). Except for hashsolo and demuxem (which doesn’t work anyway), all modules require 10X Genomics input. If we want to support other formats, I could write a module to convert AnnData or CSV into a 10X Genomics-compatible format.

I still don't think that it needs to be 10X Genomics input. My guess is that it needs to be in mtx format but this doesn't mean 10X only. Let's not overcomplicate and I'm fine with any lingering limitations we might have.

I would recommend to get the repo to the nf-core organization first (and maybe even do the 1.0 release) before adding any additional features - otherwise there will be a 1000 other ideas before this is done (this is what usually happens to me and it's not great)

I still don't think that it needs to be 10X Genomics input. My guess is that it needs to be in mtx format but this doesn't mean 10X only.

If you mean just a

.mtx

file with "mtx format" it won't work because 4/7 use

Read10X()

in their R script. This function only reads "Directory containing the matrix.mtx, genes.tsv, and barcodes.tsv files provided by 10X" . I also tested it:

> counts <- Read10X(data.dir = "/test_data/hto/matrix.mtx")
Fehler in Read10X(data.dir = "/test_data/hto/matrix.mtx") : 
 Directory provided does not exist
> counts <- Read10X(data.dir = "/test_data/hto")

hto
├── barcodes.tsv
├── genes.tsv
└── matrix.mtx