Automation of the Charmm General Force Field for Drug-Like Molecules

Abstract

The CHARMM Force Field[1] is widely used for the study of biological macromolecules. Recently, we released the CHARMM General Force Field (CGenFF) which allows simulating drug-like molecules in an environment represented by the CHARMM biomolecular force field.[2] To facilitate the use of CGenFF, a computer program was developed that automatically assigns atom types, bonded parameters and charges to any organic molecule. The atom typing is rule-based and programmable, making it easy to update the atom typing scheme as the force field grows. Assignment of bonded parameters is based on substituting atom types in the definition of the parameter, and returns a “penalty score” as a measure for the accuracy of the approximation. Charges are assigned using an extended bond-charge increment scheme that should be able to capture short- and medium-range inductive and mesomeric effects.[3] This functionality is accessible on the web, which makes it possible to apply CGenFF on arbitrary molecules in a matter of seconds.[4] Although the program should ideally accept any molecule, the quality of the resulting parameters (as indicated by their “penalty scores”) varies depending on whether a similar compound has been parametrized before. Therefore, an effort has been started to build a computational engine with the capability of automatically validating and/or optimizing parameters for drug-like model compounds of the user's interest in the framework of CGenFF, thus providing high-quality force field parameters for computer-aided drug design.[4]1. A. D. MacKerell et al., J. Phys. Chem. B 1998, 102, 3586-36162. K. Vanommeslaeghe et al., J. Comput. Chem. 2010, 31, 671-6903. K. Vanommeslaeghe et al., in preparation4. NSF Award #0823198; https://www.paramchem.org/