Hello,
I'm running an analysis pipeline, and I'm seeking some clarification on how glmGamPoi handles offset matrixes (and edgeR, as it seems to be similar in this case).
I suspect I'm interpreting this correctly, but love a sanity check.
Effectively my question is: Does glmGamPoi use an offset matrix that's added element-wise for each observation?
From the glmGamPoi paper: "Step 6: calculate the mean matrix M = exp(B X^T + D), where B is the coefficient matrix, X is the design matrix, and D is the offset matrix.
From the glmGamPoi R code for the glm_gp function: constant offset in the model in addition to log(size_factors). It can either be a single number, a vector of length ncol(data) or a matrix with the same dimensions as dim(data)
Also reading the code the for "combine_size_factors_and_offset" function, my read is that glmGamPoi makes a genesXsamples matrix of offset values by combining any provided offset matrix or vector with a vector of log(sample_size).
It's not explicitly stated that the offset matrix is used this way, and I want to see if anyone had some insight into whether or not I'm reading this implementation correctly.
Thank you
You can ask over at support.bioconductor.org if you don't get a response here. Both the lead develoer and others that will know are active over there (here as well but not as frequent).