ABSTRACT
Computer simulations give insights into the structures, transitions, and mechanisms of proteins, in terms of the underlying atomic interactions. Simulations can complement experimental data, by leveraging physical principles, by adding the time dependences, and by giving detailed narratives of molecular processes. Experiments are often coarse-grained over time and/or space, and do not provide details atom-by-atom and nanosecond-by-nanosecond. Computer simulations can fill in those details. Knowing the details is like watching a whole movie, rather than just seeing the opening and closing scenes. Simulations can also help with the design of proteins or drugs; they can help you refine protein structures; and they are useful for making movies or visualizations of mechanisms. Reviews are given by the group of Schulten (Perilla et al.) [1], Karplus and Kuriyan [2], and Dror et al. [3].
Computer simulations must search and sample protein motions and conformations. You need to explore different conformations in order to model dynamical processes and to seek stable states, having low free energy. Here’s a metaphor. Suppose you want to find the main walking path or deepest valley on a mountainside. It’s not likely that you’ll find either of them by a static snapshot or by random meandering. However, there are principled strategies that help you find deep valleys or important pathways efficiently, without exhaustively searching everywhere.1
