Another axis is batch structure. Different batches can be linked to different experimental conditions, treatments, or perturbations, and show substantial variation. Accordingly, keju further improves sensitivity through RNA batch-specific uncertainty estimation. (4/n)