(712a) Critical Evaluation of Implicit Solvent Hydration Free Energies from EEF1, Absinth, and GB/SA Versus Explicit Solvent Molecular-QCT Calculations

Implicit solvent models enable long time simulations and thereby allow one to investigate the structure and phase behavior of polypeptides. However, the ability of these models to predict polypeptide hydration thermodynamics remains unexplored, likely because of the challenge in calculating hydration thermodynamics of polypeptides in all-atom simulations. The molecular quasichemical theory (mQCT) of solutions has addressed this challenge and makes available a set of rigorously validated hydration free energies of polypeptides, including archetypes of intrinsically disordered peptides. On this basis, we study popular implicit solvent models effective energy function (EEF1), its descendent (ABSINTH), and the generalized-Born surface accessible (GB/SA). Qualitatively, GB/SA consistently makes more accurate distinctions between structures compared to ABSINTH and EEF1. Quantitatively, GB/SA is better than both ABSINTH and EEF1. Our results reaffirm the understanding that a more explicit incorporation of electrostatics improves the specificity of a model. We also discuss correlation effects between groups that are missing in the additive EEF1 and ABSINTH approaches.