Developing molecular docking methods to model protein interactions

Abstract

[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT AUTHOR'S REQUEST.] In many cellular processes, proteins carry out their functions through interacting with other molecules, such as small compounds (i.e. ligands), peptides or other proteins. Computational modeling of these interactions with molecular docking has great applications in mechanistic investigation of protein binding and therapeutic development. This PhD dissertation presents my studies on developing molecular docking methods to model protein-ligand interactions and protein-peptide interactions. The studies include the improvement of statistical potential-based scoring functions for evaluating protein-ligand interactions, the development of a docking-based method for predicting peptide binding sites on protein surfaces, and the development of a fully blind docking method for predicting protein-peptide complex structures. The scoring functions are implemented in our docking software, and the peptide binding methods are implemented in web servers. Finally, for future work, the existing methods for modeling protein flexibility in molecular docking are reviewed in detail.