@simocristea For sure. Still, effect size is important, and there are ways to use it. I played around with LFC shrinking and sorting by posterior LFC (drafted, never published), there is the old TREAT (null is LFC < X with non-zero X), and newer topcon
This is why I prefer the topconfects package to estimate a CONfident efFECT (confect) size for each gene by ~essentially running TREAT (limma null hypothesis test relative to a threshold) at all possible thresholds. https://t.co/qFcSngqdNg
RT @MonashBioinfo: Tool Technique Tuesday. Do you want TOP results by CONfident EfFECT sizes? Then Topconfects is for you. Another tool by…
RT @MonashBioinfo: Tool Technique Tuesday. Do you want TOP results by CONfident EfFECT sizes? Then Topconfects is for you. Another tool by…