LAUSR

Significance Proteins that undergo phase separation promote biomolecular condensate formation and play a significant role in many biological processes. We divided these proteins into two categories according to their underlying driving force when forming condensates: self-assembling proteins, which interact with the same protein species, and partner-dependent proteins, which interact with different biomolecule species. Most of the current computational tools preferentially predict self-assembling proteins and perform poorly in screening partner-dependent proteins. We thus built machine-learning models to predict the two protein categories separately. Further validation on the condensate proteome revealed that partner-dependent proteins are widespread in cells. We also developed a web server that integrates multiple phase-separation predictors, providing a convenient way for biologists to discover candidate phase-separating proteins.