aertslab / scenicplus

SCENIC+ is a python package to build gene regulatory networks (GRNs) using combined or separate single-cell gene expression (scRNA-seq) and single-cell chromatin accessibility (scATAC-seq) data.
Other
186 stars 29 forks source link

Avoid discarding cells if one modality failed. #421

Open ktroule opened 5 months ago

ktroule commented 5 months ago

Hi.

How do you overcome the issue of discarding cells (multiome data) for the peaks calling if a cell has been discarded/unnannotated during the analysis of the nuclei? If I'm not wrong (from the tutorials), if a cells is discarded on the nuclei modality the annotation of this cells will not be available for the peaks calling of the ATAC and hence will be discarded from that analysis even if the ATAC data is of good quality.

I want to avoid discarding a cell just because one of the two modalities failed. I can do the peaks calling outside os scenicplus, but wondering if this is can be handled within scenicplus?

PD: One possible option is to provide a dummy label to all unassigned nuclei (i.e unassigned).

Tanks for your time.