Will Macnair

Hey, thanks @jalvesaq for the awesome package and others for making the nice colour schemes easily accessible. I'm trying to use the gruvbox scheme from @sjlva, however I'm getting strange...

Thanks for the pointer, I'll look into that.

One thought is whether it could be related to offset terms. In `insight:::.variance_distributional`, a warning is thrown if `exp(null_fixef)` is too small. This makes sense if there is no offset,...

Hi @bwiernik, thanks very much for the comments and resources. Regarding model choice, something like a mixed negative binomial model is clearly the right choice for this kind of data;...

Also a brief note on your zero-inflated model - you get R2 values of around 13%: ```r fit_glmm2 = glmmTMB::glmmTMB(counts ~ celltype + (1|individual), offset = log(libsize) - log(1e6), ziformula...

I'm not sure I get why a more complex model wouldn't give a better fit here. The set of models without zero-inflation is a strict subset of the models with...

I think I'm having a similar issue, but when calling `r2_nakagawa` from inside a function (i.e. a more general issue than `map`). Maybe `stats::update` expects to be called in the...

Hi both, thanks for your detailed responses. I need to read them carefully, so it will take me a little while to respond to the tricky bits. On the easy...

A bit of further analysis, looking at all of our samples. It seems that we consistently get ~5% of permit-list barcodes ending up with 0 genes (ignoring two samples with...

Hi both, thanks again for taking the time! Some answers / responses, slightly reordered to help me make sense of new information: > Are any of these samples sharable? >...