PREDICTIVE QSAR MODELING

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The development of highly efficient and accurate approaches to virtual screening continues to represent a formidable challenge in the field of computational drug discovery. This Competitive Revision application is submitted in response to the NOT-OD-09-058 titled: NIH Announces the Availability of Recovery Act Funds for Competitive Revision Applications. The parent grant of this application is focused on enabling significant enhancements in predictive QSAR modeling technologies and their application to identifying computational hits in large chemical databases. In the course of studies enabled by the currently funded project we have realized that some of the QSAR modeling approaches could be actually extended towards a complimentary filed of structure based drug discovery. Building upon our experience in cheminformatics and QSAR modeling, this proposal aims to develop novel computationally efficient cheminformatics approaches applied to structure based virtual screening and ligand pose scorings. Furthermore, in collaboration with a specialist in structure based modeling approaches, Dr. N. Dokholyan, we endeavor to combine these cheminformatics-inspired methodologies with empirical forcefield based approaches towards the selection of progressively smaller subsets of target-specific high-affinity ligands from exceedingly large chemical libraries available today for experimental biological screening. The ultimate goal of our hybrid methodology is to arrive at a small set of high-affinity computational hits in receptor-bound conformations that could be validated experimentally. This goal will be achieved by structuring the proposed studies around the following three Specific Aims: 1) Develop novel highly efficient structure based cheminformatics approaches to virtual screening; 2) Develop statistical structure based pose scoring and binding functions using novel geometrical chemical descriptors of ligand-receptor complexes; 3) Develop hybrid approaches integrating statistical and empirical scoring functions for pose refinement and binding affinity prediction. We expect that the implementation of the methods advanced in this application will significantly enhance the repertoire, efficiency, and availability of advanced computational tools for computer aided drug discovery. PUBLIC HEALTH RELEVANCE: Computer-aided drug discovery methods play significant role in facilitating the selection of most viable lead compounds for subsequent pharmaceutical development. It is critical to develop most computationally efficient and theoretically robust approaches to ensure the usefulness of computational predictions of hit and lead chemical structures. This proposal advances the efficient and robust computational workflow for structure based virtual screening of available compound collections to arrive at a small number of reliable and experimentally testable candidate molecules with high predicted binding affinity to their biological targets.