Harnessing a vast natural dataset to comprehensively characterize biological protein condensation

Biomolecular condensation is a physical process by which proteins form macromolecular assemblies in cells. Intrinsically disordered regions (IDRs) often drive the formation of biological condensates. Proteins that undergo biomolecular condensation are overrepresented in human pathology. For example, we recently showed that mutations that cause human diseases including intellectual disabilities, cancers, and neurodegeneration are overrepresented in proteins predicted to phase-separate, and that mutations in IDRs predicted to drive protein condensation are more likely to be pathogenic than those in IDRs not predicted to phase separate. Multiple predictors and physical frameworks are currently used to identify which, and how, IDRs drive biomolecular condensation. Current models of the molecular mechanisms and biological roles of protein condensation are limited by the small number of proteins whose condensation has been studied experimentally. We are combining computational analysis of evolutionarily related sequences with medium-throughput experimental assays to expand our understanding of IDR-driven condensation. We aim to characterize molecular mechanisms & biological roles of protein condensation and improve sequence-based prediction of intrinsically disordered protein condensation.