Date of Award
Doctor of Philosophy (PhD)
Molecular dynamics simulations are a powerful tool to explore conformational landscapes, though limitations in computational hardware commonly thwart observation of biologically relevant events. Since highly specialized or massively parallelized distributed supercomputers are not available to most scientists, there is a strong need for methods that can access long timescale phenomena using commodity hardware. In this thesis, I present the goal-oriented sampling method, Fluctuation Amplification of Specific Traits (FAST), that takes advantage of Markov state models (MSMs) to adaptively explore conformational space using equilibrium-based simulations. This method follows gradients in conformational space to quickly explore relevant conformational transitions with orders of magnitude less aggregate simulation time than traditional simulations. Since each of the individual simulations are at equilibrium, all of the thermodynamics and kinetics in the final MSM are preserved. Here, I first describe the FAST method then demonstrate that it can be used for a variety of tasks, from folding proteins to finding cryptic pockets. Next, I validate that FAST discovers appropriate transition pathways between states. Lastly, I apply FAST in detailing the mechanism of stabilization for a clinically relevant mutation in TEM-1 β-lactamase. This mechanistic understanding is then used to design other stabilizing mutations, which are all supported experimentally.
Chair and Committee
Gregory R. Bowman
Alexander B. Barnes, Timothy M. Lohman, Rohit V. Pappu, Jay W. Ponder,
Zimmerman, Maxwell Isaac, "FAST-Forward Protein Folding and Design: Development, Analysis, and Applications of the FAST Sampling Algorithm" (2019). Arts & Sciences Electronic Theses and Dissertations. 1974.