fitting model without using gene sets

[willtownes]

I was wondering if it is possible to fit the fscLVM model without using gene-sets? Would I just specify a single pseudo-annotation with all genes pointing to it?

[willtownes]

I figured this out myself. Just create a single fake pathway annotation containing all genes. This is equivalent to there being an extra dense factor in the model. So, there's no way to fit it with only sparse factors, but if you are ok with having at least one dense factor and however many sparse factors desired, it works well.