Computational Methods for Conformational Sampling of Biomolecules

Proteins play a fundamental role in virtually every process within living organisms. For example, some proteins act as enzymes, catalyzing a wide range of reactions necessary for life, others mediate the cell interaction with the surrounding environment and still others have regulatory functions. Recent studies have demonstrated that the specificity of the biological function and activity of proteins is intimately linked to their structural and dynamical properties. In principle, these properties can be calculated using computational techniques. However, most structural transitions of biological relevance occur on time-scales inaccessible to current methodologies due to prohibitive computational costs. In this dissertation I present a number of new methodological improvements for calculating structural and dynamical properties of proteins at long time-scales. First of all, we have developed a new mathematical approach to a classic geometrical problem in protein simulations, and demonstrated its superiority compared to existing approaches. Secondly, we have constructed a more accurate implicit model of the aqueous environment, which is of fundamental importance in protein chemistry. This model is computationally much faster than models where water molecules are represented explicitly. Finally, in collaboration with the group of structural bioinformatics at the Department of Biology (KU), we have applied these techniques in the context of modeling of protein structure and flexibility from low-resolution data.