Hello,

For one comparison of my dataset, after shrinking the LFC and plotting the volcano plot, I get a set of data points which are non significant but they do not get shrunken. The same is observed in the MA plot.

I was wondering what the reason could be here?

Thanks!

This has been posted before, but maybe hard to find: apeglm shrinkage is often more conservative than the point null hypothesis p-values.

Two ways to square this is 1) to perform inference on the posterior LFC, by setting svalue=TRUE, or 2) use lfcThreshold in results() to do a bit more than just point null testing.