Use in RNA-Seq + gene expression?

[gerverska]

Hello everyone,

Given the generality of gllvm, I'm wondering if any current users are looking at (or are currently) applying gllvm to gene expression analyses. I was only recently reminded to ask this question after coming across another paper in my field where DESeq2 is used for differential abundance tests.

At this point, the DESeq2 pipeline is pretty commonplace for studies that examine gene expression in some way, but I don't see why practitioners wouldn't prefer gllvm over DESeq2. In addition to more flexible model design (including random effects, row effects, and offsets), residuals investigation, and obviously latent variable ordinations, gllvm has more to offer.

Another advantage I thought gllvm had over DESeq2 was that features were jointly modeled, but the DESeq2 paper seems to indicate that there is a process for sharing information across genes when estimating dispersion parameters. However, it's not apparent how similar these joint modeling processes are.

I'm curious to hear if anyone has had any experience trying both tools out. I think it'd also be fun and interesting to run them side by side on the same test dataset. It does seem like the field should shift to gllvm for all the added features, though.