Extending the Applications of Boxed Molecular Dynamics: from Simulating Atomic Force Microscopy Experiments to Sampling Trajectories from Virtual Reality

This thesis presents Boxed Molecular Dynamics (BXD) as a useful method of accelerated sampling. It can be used to circumnavigate the rare event problem conduct simulations on extremely long timescales inaccessible to other forms of molecular dynamics, as well as to tackle complex problems for which a simple reaction coordinate cannot be defined and must be described in multidimensional collective variable space.
The BXD method is discussed in both its most primitive one-dimensional form as well as after extension to multidimensional collective variable space. This is followed by a presentation of two new developments to the BXD method, which advance the scope of BXD simulations.
1) Using a one-dimensional reaction coordinate, protein unfolding Atomic Force Microscopy experiments are simulated over a range of pulling velocities. Modifications to the results of unbiased BXD simulations combined with solution of the kinetic master equation allows Atomic Force Microscopy to be modelled at pulling speeds inaccessible to other forms of simulations, helping bridge the gap between experimental and computational methods.
2) A new simulation pipeline for generating free energy surfaces is introduced in which trajectories from virtual reality are used to both define a set of collective variables for the system and as a path for the dynamics to follow. Results for three test systems, each presenting their own unique challenges are reported as a proof of concept for the method. Lastly, as a final validity check a free energy profile is generated for the unfolding of I27 and compared to a previously published version taken from BXD simulations in CHARMM.