Modified amino acids (MAAs) play a pivotal role in studying protein dynamics. Spectroscopy experiments use MAAs to probe the dynamics and structure of biological macromolecules. Molecular Dynamics can be complementary to these experiments, providing a model for the atomic details underlying the signal obtained. However, to make structural predictions, or to study structural perturbations introduced by the MAAs, the energy function (force field) that describes the conformational changes of the MAA must be accurate.A pivotal part of a Molecular Mechanics (MM) forcefield is the dihedral energy function, which corrects for differences between the energy profiles of the model and true system. The correct relative energy is obtained by fitting dihedral parameters to improve the match to more accurate reference data, such as quantum mechanics(QM) energies. Two problems encountered when fitting MM to QM are termed the origin- and the magnitude-problem. The magnitude-problem refers to the differences in the magnitude of the QM and MM energies, preventing the use of absolute energies for fitting; hence relying on relative energies and the need for an origin point. The origin-problem refers to the selection of a conformation on the potential energy landscape as the origin point to match MM energies to QM energies. Our approach to both problems utilizes each conformation of the molecule as the origin point in a pairwise manner.The relative energy error between each pair is calculated, and our objective function minimizes the average absolute error (AAE) between the total pairs of conformations.o overcome the increase in computations for each pair and help optimize the search for parameters in a discrete search space, we developed a GPU accelerated genetic algorithm, RAGTAG, for fitting dihedral parameters to any source of reference energies. First, we describe the usability of RAGTAG and use few examples of using it as proof of concept. Next, we calculate reference energies (MP2/631++G**) for 6 MAAs and fit accurate dihedral parameters that best describe their conformational landscapes.We used the newly parameterized MAAs to study the dynamics of T4-Lysozyme, HIV-1 pr, and IAPP. The results showed good agreement with experimental data from EPR and FRET. Overall, we present a new MM parameter fitting tool, and an example application of 6 MAAs: acetylated-lys, cyano-phe, azido-phe, MTSL, tyr-, and seleno-met. The AAE from the RAGTAG Fit ranged from 0.89 – 1.16 kcal/mol.

GPU-based force field fitting program and application to new parameters for modified amino acids

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.