Understanding protein dynamics from first principles has been a grand challenge in the computational chemistry community. Here we present our ab initio molecular dynamics simulations of representative polypeptide and mini-proteins, including Lasso peptide benenodin-1, Thuricin CD, and calcium channel blocker omega conotoxin MVIIA. These simulations reveal the time-resolved behaviors of proteins from a fully quantum mechanical level, showing how the large-scale electronic structures evolve and equilibrate, and how charge transfer and dynamics can occur in the long time-scale simulations. Such simulations are only made possible by recent advances in hardware-accelerated (i.e., graphical processing units, or GPUs) electronic structure methods that now make the fully first-principles simulation of larger systems (i.e., 300-500 atoms) amenable to first-principles dynamics. In analogy to dynamic coupling or mutual information widely employed in the classical molecular dynamics community, these observations allow us to identify and characterize the coupling and fluctuation of electronic properties (e.g., partial charges or local dipoles) that are driven by spatial contact or electrostatic interactions. Such observations could motivate development of improved force field treatments, systematic QM/MM methodologies, or machine learning-based approaches that identify when improved force fields or full-QM simulation are essential to model charge transfer and polarization more fully.

Revealing the quantum mechanical nature of proteins through <i>ab initio</i> molecular dynamics

256th National Meeting (2018)

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

Revealing the quantum mechanical nature of proteins through ab initio molecular dynamics