The FuzDrop method predicts the probability of proteins to undergo liquid-liquid phase separation. It performs a sequence-based identification of both droplet-promoting regions and of aggregation-promoting regions within droplets. These features can be visualised on existing structures or those predicted by AlphaFold.