An accurate description of electronic polarization is fundamental to modeling protein interactions and dynamics. Despite its importance, polarization is frequently omitted in classical force fields, while existing polarizable models are hindered by parameterization complexity, high computational cost, or limited resolution. We present a novel, efficient method for calculating protein polarization energies. Our method computes local electric fields at atomic sites, decomposes them into bond-parallel and perpendicular components, and uses environment-specific, anisotropic atomic polarizabilities fitted to accurate quantum-chemical calculations. The model accurately reproduces quantum-mechanical polarization energies for amino-acid monomers, dimers, trimers, and small proteins. Its computational efficiency enables application to large, explicitly solvated proteins, achieving excellent agreement with benchmark data at a minimal computational cost. We demonstrate its scalability by applying it to large, solvated proteins, providing a practical path for incorporating polarization into biomolecular simulations.
Ren et al. (Thu,) studied this question.