Development of a Polarizable Force Field for Macromolecules Based on the Classical Drude Oscillator

Accurate potential functions based on simple and computationally tractable functional forms are essential for meaningful molecular dynamics simulation studies of biomolecular systems. Towards this need we have developed a polarizable force field based on the classical Drude oscillator model. Built on a foundation of parameters optimized targeting a range of experimental and quantum mechanical (QM) data for small molecules representative of macromolecules, force fields for proteins, lipids, nucleic acids and carbohydrates have been developed. Extension of the polarizable model from small molecules to macromolecules required additional optimization of the electrostatic parameters in the context of the biopolymers, iterative optimization of the torsion potentials targeting both QM and experimental data and the inclusion of pair specific Lennard-Jones parameters. Examples from optimization of the polarizable model for proteins and nucleic acids will be presented, including results from explicit solvent simulations of 100 ns and longer.