Performance comparison of Poisson–Boltzmann equation solvers DelPhi and PBSA in calculation of electrostatic solvation energies

Abstract

Many Poisson–Boltzmann equation (PBE) solvers have been developed to calculate electrostatic energy of biomolecules, and each PBE solver has its advantages. In this study two PBE solvers — DelPhi and PBSA module in AMBER — are compared in terms of calculation results, convergence and program-running time by calculating the electrostatic solvation energy for a test set composed of 4 Kirkwood models and another test set of 25 protein structures. The protein structures were pretreated by AMBER and AMBER99SB force field parameters were used in all calculations. It is found that (i) At fine grids, both PBE solvers can produce accurate results on test set 1 and consistent results with each other on test set 2, with differences between each other varying from several to ~ 10 kcal/mol. (ii) Under convergence criterion "absolute value of relative error is less than or equal to 2% (|RE| ≤ 2%)", both PBE solvers need very fine grids to produce convergent results on small and complex Kirkwood models, while grid spacings of ≤ 0.5 Å–0.6 Å are well enough for them to achieve good convergent results on various molecular structures. We recommend users to adopt such grid spacing in using of these PBE solvers so as to get good enough convergent results. (iii) In terms of time consumption, DelPhi appears to be more time-saving than PBSA. In summary, according to our comparison, DelPhi and PBSA are paralleled good PBE solvers. The convergence of PBSA is a little better than DelPhi, while DelPhi exceeds PBSA in running speed.