Computing ensemble properties for small solute molecules and poly-peptides with MP2, LMP2, and density functional theory.

by Yuan, Ying; Zheng, Dong; Wang, Feng

Potential energy surfaces based on MP2, LMP2, and DFT are mapped on simple energy expressions using adaptive force matching. Such energy expressions are specifically tailored for capturing the ref. PESes in a condensed phase environment for the thermodn. condition of interest. Tests on several small solute mols. demonstrate that the models created predict ensemble properties, such as hydration free energies, enthalpies of vaporization, diffusion consts., in good agreement with expts. Predictions based on MP2 and B3LYP were able to reach chem. accuracy. For menthol, an outsider exptl. value were identified with the help of B3LYP based computational predictions. Alanine and glycine conformation distributions were predicted with various functionals, such as PBE, BP86, and B3LYP with both the aug-cc-pVDZ and def2-TZVP basis sets. When validated with exptl. NMR scalar coupling data, the DFT based models perform similarly or better than some of the best traditional protein force fields for these peptides. Of the ref. method tested, B3LYP/D3(BJ) with the def2-TZVP basis set gave the best overall performance for glycine.