drisso
commented
6 years ago Hi,
I'm not sure what you mean by "genes that contribute the most to the variability".
You can think of our model as a factor analysis model, so the alpha parameter will give you a sort of weight on how much each gene contributes to each factor (if that's your question). You can access the alpha parameter by using the getAlpha_mu() or getAlpha_pi() functions, depending on whether you want to look at the mean or at the zero inflation probability.
Does this answer your question?
bio-la
Hi Davide, what would be a measure of gene loadings in the Zinb model when used as dimensionality reduction e.g. PCA - does it make sense to rank the 'genes that contribute the most' to the variability explained by the K component estimated (and have a measure of relevance of the K)? I am currently checking the R^2 between the known covariates/batch effects and the estimated Ks, to select them for downstream analyses, as it seems it's what you suggest to do in the paper anyway. Does this make sense? Thanks.