Mechanistic insights on important biomolecules derived using simple dynamics models from extending the reach of elastic network modeling
The dynamics of biomolecules are important for carrying out their biologic functions, but these remain difficult to probe in detail experimentally, so that their accurate computational evaluation is an important field of ongoing study. Critical questions remain open such as what are the importance of individual interactions within a structure, the composition of denatured states and equilibrium native ensembles, as well as the role and conservation of flexibility in functional dynamics. The tools of Molecular Dynamics, Monte Carlo simulation, and Normal Mode Analysis coupled with knowledge-based approaches represent the mainstay of computational approaches used in this field.
The primary focus of this dissertation is to explore the functional dynamics of important biomolecules while extending the utility of Normal Mode Analysis using Elastic Network Models through the application of novel analysis methods. Many of these techniques have been made available to the scientific community through the software tool MAVEN which integrates and automates many of the steps in model building and analysis. By utilizing these tools, we have discerned structural dynamics characteristics and mechanistic behaviors of antibodies, ribosomes, telomerase, and efflux systems. Modes from multiple Anisotropic Network Models capture collective as well as local motions which accurately describe a large set of experimental tRNA structures. Mechanistic understanding of biomolecular motion can aid in the understanding of physiology, disease states, and our ability to engineer new structures with novel functions.
The ability to distinguish native-like structures from a set of computational predictions is important not only in structure prediction, but also in molecular docking and for predicting conformational changes. We propose a new algorithm for evaluating the entropy of motion of biomolecules, showing that it leads to enhanced discrimination between native-like and non-native-like models in both structure predictions and protein-protein docking. Our findings indicate that the shape of a protein or complex contains sufficient information to distinguish it from poorer quality predictions. Graph theoretical approaches have also been employed to investigate the connectedness of the protein structure universe, showing that the modularity of protein domain architecture is of fundamental importance for future improvements in structure matching. All of the studies herein impact our understanding of protein domain evolution and modification.