Hi,

In a recently published Bioconductor vignette of apeglm: https://bioconductor.org/packages/devel/bioc/vignettes/apeglm/inst/doc/apeglm.html#Modelingratiosof_counts,

it detail a method of modelling allelic imbalance or ASE using a model however I have some question about the steps to complexify the base model. Essentially, I have 4 hybrid individuals, with 2 counts (2 species), in 3 conditions within a continuum. I would like to test the following design ~condition + species + condition:species.

I imagine within the first step : The model matrix would be shifted to x<- model.matrix(~Condition + species + Condition:species)

and the coef would of course be ajusted to to column within this model which we want to test.

The point which I have the most trouble is when testing the interaction: As the threshold argument would not necessarily be 0.5 ?

fit2 <- apeglm(Y=ase.cts, x=x, log.lik=NULL, param=param, coef=coef, mle=mle, threshold=0.5, log.link=FALSE, method="betabinCR")

Is there a way through apeglm to test this interaction ?

Thank you.

The threshold is whatever you like (as in gene-level counts). Remember, the link with a beta-binomial is now logistic, so a coefficient of 0 represents the alleles being equally expressed. To figure out what threshold you might want to use, consult the logit (as we do at the very end of the count ratios section).