Gregor Sturm

The pipeline should fail on invalid parameter combinations. One example: * it's possible to specify `--transcript_gtf`, but it has no effect on kallisto

Just one note, some time ago I created [this shared module](https://nf-co.re/modules/custom_gtffilter/) for filtering the GTF based on FASTA. Maybe it is useful for scrnaseq _Originally posted by @nictru in https://github.com/nf-core/scrnaseq/pull/464#pullrequestreview-2789928144_

Use different json schemas for different aligners

1

There's a lot of custom validation code necessary for e.g. cellranger-arc: https://github.com/nf-core/scrnaseq/pull/441/files#diff-f8dc84dd3af7df7368a655912bcfcb589853aa49a4019041f6a772d1eb3506c1R217-R252 All of this could technically be represented as a json schema, but we currently only have one universal...

Check that indices can be specified as expected

7

### Description of the bug Specifying indices should work as expected from a nf-core pipeline, i.e. * igenomes should be respected * it shall be possible to specify pre-computed indices...

### Description of feature When running cellranger multi, the pipeline should create a MuData object with all modalities instead of an AnnData object. Depends on #369

Move local modules to nf-core/modules

2

### Description of feature Some of the modules (especially those wrapping a single tool) look like they could go to nf-core/modules and be installed into the pipeline.