I recently upgraded my R to 3.1.2 and DESeq2 to 1.6.3. I re-run some of my code on a RNAseq experiment. Comparing to the old results (done in R 3.1.0 and DESeq2 v1.4.5), all the estimates (e.g. fold changes, p values) have changed. It seems that the most dramatic changes happened to those genes with large fold changes (I haven't checked carefully if these genes are those with low read counts). Could somebody give me some pointers on what exactly changed between the two versions that would cause these changes? I read through the package's news file and it wasn't obvious to me.