Elastic network models of biomolecular structure and dynamics

Macromolecules (e.g., proteins and nucleic acids) play a critical role in living cells. Molecular motions are involved in many basic functions of the cell such as catalysis, regulation, transportation, and aggregation. Comprehending such biological mechanisms may be the first step in understanding the phenomena of life. This dissertation is devoted to the study of biomolecular structure and dynamics. Many engineering analysis tools such as kinematics, vibrations, linear algebra, and statistical mechanics are adopted to solve various biological problems in this dissertation.; In a coarse-grained elastic network model, a system is represented as a network of springs connecting representative point masses. For example, only Calpha atoms in a protein are treated as point masses and spatially proximal points are assumed to be linked with linear springs. Normal mode analysis (NMA) with a simple harmonic potential function is performed to study the dynamics of a macromolecule around an equilibrium state. This is computationally more efficient than conventional approaches such as molecular dynamics (MD) or even NMA using full-atom empirical potentials.; We develop elastic network interpolation (ENI) which is a purely geometry-based technique. The key idea is to uniformly interpolate the distances in two different conformations within the context of the elastic network model. ENI generates a feasible reaction pathway between two different conformations. It is suitable to describe the global motions of complex systems of small proteins or single proteins having more than several thousand residues within reasonable time on a desktop PC. In instances when only partial conformational data are obtained from experiments, ENI can be used to incorporate those incomplete information in computer simulations. ENI is also used to interpret massive amounts of MD data by finding essential pathways.; ENI has been modified to save computing power in different systems. When the motions of part of a system look like rigid-body motions, the system can be represented by rigid-clusters (called rigid-cluster ENI). If a system consists of repeated units such as virus capsids, only one repeated unit and its surrounding conditions need to be considered (called symmetry-constrained ENI). Several examples validate both modified ENI methods.; Consequently, ENI may serve as a paradigm for reduced-DOF dynamic simulations of large macromolecules as well as a method for the reduced-parameter interpretation of MD data. Good agreement with experimental data validates elastic network models as a tool for the study of biomolecular structure and dynamics.