Assessing multi-eGO Predictions of PDZ2-Peptide Binding across Mutations

Multi-eGO predicts the equilibrium dissociation constant for the PDZ2–EQVTAV complex within expected uncertainty, while accelerating both binding and unbinding by several orders of magnitude.

• Why it matters: Protein–peptide binding changes from mutations are hard to simulate because both conformational sampling and binding kinetics are computationally expensive.
• What they did: Using PDZ2 from protein tyrosine phosphatase basophil-like bound to EQVTAV, the model was calibrated on explicit-solvent complex simulations and then applied to conservative PDZ2 and peptide mutations without per-mutant training.
• The result: Predicted PDZ2 mutant dissociation constants match experiment (Pearson r = 0.97), and maximum-caliber reweighting attributes most kinetic mismatch to a single global acceleration factor, leaving system-specific rate errors recoverable by reweighting.

AI summary of the abstract · may contain errors · not medical advice · read the paper.