www.sciencedirect.com/science/arti...

@jemmageoghegan.bsky.social I just took a quick look, I'm pleasantly suprised by their negative controls. But the cross validation is pretty poor. They really should have reduced sequence similarity between the training and test sets.