If one imagines activity-directed synthesis to resemble a game of Battleship, then in silico data mining is likesonar: rather than blasting through all chemical space near preliminary hits until tangible patterns emerge, onecan mine the wealth of preliminary data to detect key underlying trends and target one's chemistry accordingly.Herein we thus propose to apply a series of computational protocols to efficient delivery of chemical insight thatwill guide targeted synthesis of hit analogs with elevated prospects for achieving probe status. Our overarchingobjective is a seamless IT pipeline that acquires, analyzes, stores and delivers all information relevant toscientific function of this Specialized Chemistry Center (SCC), specifically focusing on delivering:1. a robust, efficient and secure information management environment that enables assimilation of all dataand metadata associated with a given screen into our own local databases in a format suitable foranalysis and internal reference and reporting of resulting analyses, data and metadata in the formatsrequired by the synthesis core, the originating screening center and the MLPCN program,2. an array of specialized in silico screening mechanisms that permit (a) facile characterization ofbioactive clusters within the preliminary screening set, (b) identification of subsets of large existingcompound collections that physicochemically overlap with such promising regions of chemistry space,and (c) intuition of novel chemistries that stand to augment and potentially improve upon existingbioactives,3. highly insightful quantitative structure-activity relationship (QSAR) models for potent families ofbioactives that illuminate key structural variants with optimal prospects for meeting viable probe criteria,and4. reliable in silico prescreens for compound solubility or other practical issues that should be gauged priorto compound acquisition or synthesis.Our access to a wealth of computational and support resources dedicated toward chemical librarydevelopment, our extensive experience in the application of the above methods toward probe development aspart of a CMLD program and PSL projects, and our established research focus on development of novelalgorithms that enhance the biological relevance, target-sensitivity and chemical information content ofmodeling paradigms render our team particularly well qualified to deliver these services.
Showing the most recent 10 out of 51 publications
