返回Google圖書搜尋

Development and Application of Computational Methods for Structure-based Drug Design

出版	Boston University, 2008
URL	http://books.google.com.hk/books?id=xFm9swEACAAJ&hl=&source=gbs_api

註釋Abstract: This dissertation explores the development and application of computational methods for structure-based drug design. First I apply a pre-existing computational algorithm, computational solvent mapping (CS-Map), to the identification of 'hot spots, ' sub-regions of binding pockets that are primarily responsible for the binding energy of drug-like molecules. Using an X-ray structure as input, the CS-Map algorithm determines energetically favorable binding positions of chemical groups along a protein surface. Here it is demonstrated for a variety of well-studied pharmaceutical targets that low-energy consensus binding positions of functional groups are highly predictive of hot spots. Of particular interest in this study was renin, a long-standing hypertension target. The development of the first FDA-approved renin inhibitor in early 2007 year is due to the utilization of a novel hot spot; this region is also identified using CS-Map. In a subsequent prospective study, novel hot spots are uncovered within the substrate binding region of neuraminidase, an important influenza target. The identification of two novel hot spots was accomplished using a representative ensemble of conformations derived via molecular dynamics simulations; development of inhibitors capable of binding to these novel hot spots may mitigate against drug resistance. In the second part of this dissertation I develop two computational methods for focused library design. Using the output of CS-Map, a population-based method is used to assess structure-activity relationships (SAR) for chemical groups residing within hot spot regions. Specifically, the optimization potential of hot spots as well as the affinity of specific chemical groups is evaluated using this method. As illustrated with renin, this approach yields comparable results to experimental studies, providing an accurate computational alternative to costly experiments. Last, I describe an information-theoretic algorithm for choosing representative subsets from chemical libraries based on chemical properties, allowing chemists to create maximally diverse screening libraries of reduced size for synthesis.