Hit discovery by high-throughput screening (HTS) carries huge burdens of time and resources. 2D Profile-QSAR and 3D Surrogate Autoshim are two experimentally parameterized, protein family-based, virtual screening methods, which harness the power of chemogenomics to create more efficient and predictive in silico tools. The methods have been applied in multiple real life kinase projects, and have shown consistent prediction accuracy and selectivity, resulting in high hit-rates even while exploring novel chemical space. The methods have been extended beyond what HTS can address to kinase selectivity predictions for hit-to-lead optimization and library design. 2D Profile-QSAR has also been extended to cellular activity predictions for kinase inhibitors. It has proven particularly useful where disconnects exist between cellular and biochemical activity. Over 80 Profile-QSAR models covering most of the human kinome have now been generated on experimental assay data. A kernel-type method interpolates from these models to generate a priori activity predictions for novel kinases with no experimental training data. The sequential floating search technique was implemented to identify a �minimal ensemble� of kinase structures for Surrogate Autoshim that halves the time for pre-docking large commercial databases. The two methods have been successfully extended to serine proteases, and evaluations against ADME targets such as hERG are now underway.

Profile-QSAR and surrogate autoshim: Utilizing chemogenomic relationships for drug discovery applications

240th National Meeting (2010)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.