Extending the MST Model to Large Biomolecular Systems: Parametrization of the ddCOSMO-MST Continuum Solvation Model

Continuum solvation models such as the polarizable continuum model and the conductor-like screening model are widely used in quantum chemistry, but their application to large biosystems is hampered by their computational cost. Here, we report the parametrization of the Miertus-Scrocco-Tomasi (MST) model for the prediction of hydration free energies of neutral and ionic molecules based on the domain decomposition formulation of COSMO (ddCOSMO), which allows a drastic reduction of the computational cost by several orders of magnitude. We also introduce several novelties in MST, like a new definition of atom types based on hybridization and an automatic setup of the cavity for charged regions. The model is parametrized at the B3LYP/6-31+G(d) and PM6 levels of theory and compared to the performance of IEFPCM/MST. Then, we demonstrate the robustness of the parametrization on the SAMPL2, SAMPL4, and C10 datasets. The ddCOSMO/MST models provide errors of ~0.8 and ~3.2 kcal/mol for neutrals and ions, respectively, showing a remarkable balanced and accurate description of cations and anions.