Development and application of novel computational approaches for computer-assisted drug design (CADD) and protein modeling

This dissertation focuses on the design, development and applications of novel computational approaches for computer-assisted drug design (CADD) and protein modeling. The CADD approaches have been developed in the areas of both ligand-based and structure-based drug design. In addition, Simplicial Neighborhood Analysis of Protein Packing (SNAPP) has been employed for the analysis of protein flexibility.; In the area of ligand-based design, I have implemented a new statistical modeling methodology based on lazy learning theory (termed ALL-QSAR). To our knowledge, it is the first application of this type of algorithm to QSAR studies. Compared with other modeling approaches, ALL-QSAR can generate models with significantly higher predictive power. This method has been successfully applied to hit identification for anticonvulsant agents via large chemical database mining. The results show that ALL-QSAR affords the detection of molecules with chemical substructures highly different from those included in the training set. Several identified hits have been confirmed experimentally. This initial success indicates that ALL-QSAR can be further exploited as a general tool for the drug design and discovery.; For structure-based drug design, a novel scoring function called ENTess has been developed for binding affinity evaluation based on Delaunay tessellation and Pauling's atomic electronegativity. This novel method was validated for a large dataset of 264 protein-ligand complexes. The resulting models can be used for fast and accurate scoring of complexes resulting from docking studies. Concurrently, I have made major contributions to the collaborative development of a novel cheminformatics approach that predicts Complementary Ligands Based on Receptor Information (CoLiBRI). I report on this fast docking program as a powerful and accurate tool to screen large databases such as WDI.; Finally, SNAPP was employed to study the stability of bacterial periplasmic binding proteins. This study indicated that SNAPP allows delineation of the role of critical residues in protein stabilization, thereby providing new testable hypotheses for rational site-directed mutagenesis in protein engineering.; In summary, I have developed several novel computational approaches for biomolecular modeling and presented examples of their application to various datasets. I expect that these new methods will find multiple uses in the areas of CADD and structural bioinformatics.