a) can one reasonably use voom/limma to handle single cell RNA-seq data, which are zero-dominated? Naive application seems to yield reasonable results.
b) it seems more appropriate to try to model the process as a hurdle (logit zero counts, negative binomial finite counts) or zero -dominated negative binomial process, however the pscl package seems to not really handle a full count *matrix*. If anyone has a small working example of hurdle pscl functions in this context that'd be great.
thanks to all,