Physics-based atomistic simulations of biomolecules offer a range of testable observables, providing critical mechanistic insights that are largely inaccessible to experiment. However challenging processes, such as order-disorder transitions, ionic interactions, and interface events near biomembrane, are difficult to model quantitatively with existing approaches. The difficulty comes from the requirement of accurate modeling of electrostatic and polarization effects in different structural states and different solvent phases. This is not easily achievable if we demand that our atomistic model is efficient enough for typical molecular processes. Our central hypothesis to address the accuracy issue is that biomolecules in different solvent phases and in different structural states can only be modeled with satisfactory transferability within polarizable electrostatics frameworks. In a major shift from existing approaches, we are exploring a polarizable Gaussian Multipole model, where all charges and multipoles are represented by Gaussian densities instead of classical points. On the other hand polarization treatments invariably reduce simulation efficiency, leading to a more pronounced efficiency issue. Thus a second major difference from existing approaches is our concurrent focus on efficiency based on a multi-scaled framework with all-atom polarizable, coarse-grained polarizable, and continuum polarizable models, which are consistent with each other. This allows them to be more easily interfaced in multi-scaled simulation methods. Our plan can be summarized in the following four areas. First we will develop a novel polarizable force field. Second we will develop and extend continuum polarizable solvent models consistent with the new polarizable force field. Third a coarse-grained polarizable force field will be developed. Finally, we will continue to apply our computational models and tools to study interesting biomedical problems that best demonstrate the potentials of the new models. We will concurrently disseminate the new models and tools to positively impact the biomedical community. Through these concerned efforts, we will offer the community a multi-scaled set of computer models to model biomolecular electrostatics and polarization for a range of interesting systems of biomedical importance.
Electrostatic and polarization effects play important roles in all basic biomolecular events and therefore are integral to the modeling of biomolecular structure and function. This project will generate a set of multi-scaled models for computational studies of biomolecules of medical importance. These models will enable discovery of knowledge on biomedical processes occurring at a wide range of spatial and temporal scales and yield unique opportunities for drug design through detailed mechanistic studies.
