MohammadHosseini, A. M.; Teimouri, H.; Gureghian, V.; Najjar, R.; Lindner, A. B.; Pandi, A.

Liquid-liquid phase separation (LLPS) is a physicochemical process through which a homogeneous liquid solution spontaneously separates into distinct liquid phases with different compositions and properties. Driven by weak multivalent interactions, LLPS in living systems allows the dynamic compartmentalization of biomolecules to regulate and enhance various cellular processes. Despite recent advances, predicting phase-separating proteins and their key LLPS-driving regions remains limited by the versatility of models. Here, we developed Phaseek, a generalizable LLPS predicting model that combines contextual sequence encoding with statistical protein graph representations. Phaseek accurately predicts diverse LLPS-prone proteins, as well as key regions and effects of point mutations within them. Proteome-wide analysis across eighteen species suggests that LLPS is evolutionary conserved and involved in multiple biological processes. Additionally, predictions by Phaseek highlight the key physicochemical and structural properties associated with LLPS. Provided as an open-access model with a user-friendly implementation, Phaseek serves as a versatile LLPS predictor for advancing fundamental and applied research in related fields.