NextGenPB: an analytically-enabled super resolution and local (de)refinement Poisson-Boltzmann Equation solver

TL;DR

NextGenPB introduces an analytical, adaptive-grid FEM solver for the Poisson-Boltzmann equation that enhances accuracy without increasing computational cost, benefiting biomolecular electrostatics modeling.

Contribution

It presents a novel analytical correction approach combined with adaptive grid de-refinement for solving the PBE efficiently and accurately.

Findings

Improved electrostatic potential and energy estimates.

Achieved high accuracy with reduced computational cost.

Validated on biomolecular benchmark systems.

Abstract

The Poisson-Boltzmann equation (PBE) is a relevant partial differential equation commonly used in biophysical applications to estimate the electrostatic energy of biomolecular systems immersed in electrolytic solutions. A conventional mean to improve the accuracy of its solution, when grid-based numerical techniques are used, consists in increasing the resolution, locally or globally. This, however, usually entails higher complexity, memory demand and computational cost. Here, we introduce NextGenPB, a linear PBE, adaptive-grid, FEM solver that leverages analytical calculations to maximize the accuracy-to-computational-cost ratio. Indeed, in NextGenPB (aka NGPB), analytical corrections at the surface of the solute enhance the solution's accuracy without requiring grid adaptation. This leads to more precise estimates of the electrostatic potential, fields, and energy at no perceptible additional cost. Also, we apply computationally efficient yet accurate boundary conditions by taking advantage of local grid de-refinement. To assess the accuracy of our methods directly, we expand the traditionally available analytical case set to many non-overlapping dielectric spheres. Then, we use an existing benchmark set of real biomolecular systems to evaluate the energy convergence concerning grid resolution. Thanks to these advances, we have improved state-of-the-art results and shown that the approach is accurate and largely scalable for modern high-performance computing architectures. Lastly, we suggest that the presented core ideas could be instrumental in improving the solution of other partial differential equations with discontinuous coefficients.