Thesis

Abstract:

Molecular dynamics (MD) simulations are used to model the dynamics of many chemical and biological reactions. However, the applicability of MD is often limited due to its inability to describe "long" time-scale processes, which is a result of small time-step required for the integration of Newton’s equations of motion. Since numerous of interesting chemical reactions encounter several metastable states separated by free energy barriers which are much higher than thermal energy, their simulation at a desired temperature could be kinetically hindered. In order to circumvent this time-scale problem and to simulate such "rare events" within computationally affordable time, several enhanced sampling methods have been introduced. Umbrella sampling (US) is one of the popular techniques where free energy of a known reaction coordinate is computed by adding several harmonic biasing potentials along a collective variable (CV). Nonetheless, higher computational time is required for performing US simulations which restricts its applicability to sample low-dimensional free energy surfaces. Another popular enhanced sampling technique is Metadynamics (MTD). This method has been shown to explore complex free energy landscapes of chemical reactions and conformational changes, and has been successfully applied to various problems in chemistry, biology, and material science. In this approach, a set of few CVs describing the process is enhanced sampled by adding history dependent Gaussian shaped biasing potential. As a result, the MTD approach self-guides the system on a high-dimensional free energy landscape which aids in predicting unprecedented minima and reaction pathways. However, as the computational time required to fill the free energy basins increases exponentially with the number of CVs, the application of MTD is often restricted to 2-3 CVs. Furthermore, MTD is inefficient for sampling the processes where the free energy basins are unbound or flat. The latter is an important point to consider, since many A+B type reactions, drug binding, etc. have such free energy landscapes. In this thesis, we have developed two new method called well-sliced MTD (WS-MTD) and temperature accelerated sliced sampling (TASS) techniques which exploit the strengths of already available methods for efficient sampling of two or more CVs simultaneously. Furthermore, we demonstrate the application of these techniques in sampling highdimensional free energy surfaces of various chemical reactions using ab initio and quantum mechanical/molecular mechanical (QM/MM) hybrid MD simulations.