avatar
Ian Sudbery @iansudbery.bsky.social

However, if this is violated, its usally believed that BH still provides an upper bound on the expected number of false discoveries, and thus is safe. This paper claims that while the above holds on average, indevidual experiments may have very large numbers of false discoveries.

aug 19, 2025, 11:07 pm • 0 0

Replies

avatar
Ian Sudbery @iansudbery.bsky.social

The proposed solutions are either generate synthetic null data to measure FDR (assumption laden), or use permutation (transcriptomics usually to low n... I think?). So I don't know what the answer is

aug 19, 2025, 11:07 pm • 0 0 • view