| Although numerous drugs currently in use were found serendipitously, more recently “rational” methods have been applied to drug discovery. This rational design has come from an understanding of how proteins and drugs interact, what the molecular bases of diseases are, and what the targets look like in three dimensions. Computational chemistry methods are indispensable in modern drug discovery. The use of computational chemistry tools can provide insight in the atomic basis for molecular recognition, and this information can guide the discovery and design of higher affinity, more specific drug molecules. In this dissertation I investigate the molecular recognition between proteins, and I assess the ability of the molecular docking program DOCK to capture molecular recognition.; The importance of protein-protein interactions in humans has become clearer since the sequencing of the human genome. Many protein-protein interactions are important drug targets, but it has been difficult to disrupt these interactions with small molecules. In Chapter I, I analyze the relationship between complex stability and ligand size, and show that it is theoretically possible to design small molecules that have as high affinity for their protein receptor as protein ligands have.; I also showed in Chapter I that different biological classes of protein-protein complexes have different maximal stabilities. This led me to investigate in Chapter II which energy term makes certain protein-protein complexes more stable, which is shown to be the van der Waals interaction energy. This kind of information can be used in the design of small molecules to disrupt protein-protein complexes.; Docking programs are widely used in the pharmaceutical industry to find new leads or optimize existing leads for drug targets. Adequate testing is critically important in the development of docking algorithms. In Chapter III I compile a high-quality test platform and show its usefulness during algorithm development.; Chapter IV is a review of the computational chemistry field, and describes different tools, and their use. We also address the physical basis for molecular recognition and how different scoring functions try to capture the physics. Furthermore, we consider how different methods should be tested, as well as future applications. |
