A Boundary-Integral Approach for the Poisson-Boltzmann Equation with Polarizable Force Fields

Area of Science:

• Computational chemistry
• Biophysics
• Molecular modeling

Background:

• Implicit-solvent models are crucial for studying biomolecular electrostatics.
• Standard force fields use point charges, while polarizable force fields offer a more realistic charge description.
• Previous attempts to couple multipolar force fields with finite-difference solvers faced challenges in mesh transfer.

Purpose of the Study:

• To implement the atomic multipole optimized energetics for biomolecular applications (AMOEBA) force field in the PyGBe boundary integral Poisson-Boltzmann solver.
• To address limitations of finite-difference methods in handling multipolar charge descriptions.
• To assess the accuracy and efficiency of polarizable force fields within a boundary integral framework for biomolecular simulations.

Main Methods:

• Implementation of the AMOEBA force field within the PyGBe boundary integral Poisson-Boltzmann solver.
• Verification and validation of the implemented software.
• Comparison with existing finite-difference solvers (APBS).
• Assessment of computational efficiency on CPU and GPU.

Main Results:

• The boundary integral approach effectively handles multipolar charge distributions without mesh transfer issues.
• Performance on CPU was comparable to volumetric methods.
• Boundary element method requires the same mesh density for both point-charge and multipolar force fields.
• The boundary integral approach yielded more consistent results for binding energy calculations compared to finite difference approximations for multipolar force fields.

Conclusions:

• The boundary integral formulation in PyGBe provides a robust and accurate method for incorporating polarizable and multipolar force fields like AMOEBA.
• This approach overcomes limitations of finite-difference methods, particularly for high-accuracy calculations like binding energy.
• The GPU implementation enhances computational efficiency for biomolecular electrostatics simulations.