Abstract: The prospects for novel drug discovery and efficient development have never looked brighter. The NIH Molecular Libraries Program, launched in 2005, has produced a resource that will revolutionize chemical genomics in a manner comparable to how the Protein Databank has fostered the field of structural biology. This keystone resource, PubChem, provides an incomparable compendium of chemical biology data to be freely mined by an informatics community that has discovered a field ripe with challenges and great rewards. A long-term culmination of the PubChem – informatics partnership could be such pharmaceutically critical capabilities as rapid identification of novel lead-hopping prospects for a given target, genome-wide assessment of potential cross-target effects (or side effects) from specific chemotypes, and systematic identification of screening collections optimized for specific target classes. Achieving such goals requires computational methods that can select specific features of targets and chemotypes that distinguish compound-target selectivity and identify other features that correlate with propensity for promiscuity. Given these general features, it is then possible to narrow a huge pharmacopoeia down to more a more narrow space of relevant compounds and/or targets from which it is possible to devise accurate targeted models that can give accurate guidance toward potent selective lead compounds. The arsenal of computational techniques for accomplishing this is blossoming into fruition in step with a concurrent aggregation of diverse chemical biology data that will enable comprehensive chemical genomics profiling. Just as the methods required to address these challenges are diverse, our lab has pursued a broad portfolio of informatics strategies, including general metrics for understanding trends that span the entire pharmacopoeia, feature selection and classification methods to ascertain general classes of compounds and targets with similar profiles, and modeling techniques to accurately characterize trends within compound and target classes.

Brief Biography of the Speaker:
Gerald Lushington graduated from the University of New Brunswick in 1995 with a degree in theoretical chemistry. After pursuing several positions as a Department of Defense contractor working on various materials science research projects plus a theoretical study of nerve agent toxicology and therapeutics, he moved to the University of Kansas where he is now the director of five research service laboratories, all focusing on the application of computational or information-intensive techniques toward the guidance of experimental pursuits in chemistry, biology and pharmacology. He is the author of nearly 120 peer-reviewed publications; he pursues structure-based drug design and chemical informatics as his primarily areas of specialization, however he supervises research in a broad range of areas in biomedical informatics and data management