Hi @Daniel Lundin, I don't know if this applies to the 16S dataset but in our case with the COX1 dataset we applied these two constrains • Selected only the ones that differ at genus level or above with confidence score at least 0.9 • Did not take in consideration genera with one sequence