An interpretable machine learning algorithm to predict disordered protein phase separation based on biophysicalinteractions

Protein phase separation is increasingly understood to be an important mechanism of biological organization and biomaterial formation. Intrinsically disordered protein regions (IDRs) are often significant drivers of protein phase separation. A number of protein phase separation prediction algorithms are available, with many specific for particular classes of proteins and others providing results that are not amenable to interpretation of contributing biophysical interactions. Here we describe LLPhyScore, a new predictor of IDR-driven phase separation, based on a broad set of physical interactions or features. LLPhyScore uses sequence-based statistics from the RCSB PDB database of folded structures for these interactions, and is trained on a manually curated set of phase separation driver proteins with different negative training sets including the PDB and human proteome. Competitive training for a variety of physical chemical interactions shows the greatest importance of solvent contacts, disorder, hydrogen bonds, pi-pi contacts, and kinked-beta structure, with electrostatics, cation-pi, and absence of helical secondary structure also contributing. LLPhyScore has strong phase separation prediction recall statistics and enables a quantitative breakdown of the contribution from each physical feature to a sequences phase separation propensity. The tool should be a valuable resource for guiding experiment and providing hypotheses for protein function in normal and pathological states, as well as for understanding how specificity emerges in defining individual biomolecular condensates. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=101 SRC="FIGDIR/small/499043v1_ufig1.gif" ALT="Figure 1"> View larger version (18K): org.highwire.dtl.DTLVardef@151f0cdorg.highwire.dtl.DTLVardef@984335org.highwire.dtl.DTLVardef@645b3dorg.highwire.dtl.DTLVardef@a22b06_HPS_FORMAT_FIGEXP M_FIG C_FIG