In chemical biology and drug discovery, promiscuity concepts have grown in importance, along with new conceptual frameworks including systems biology and systems pharmacology. Accordingly, one-dimensional notions of promiscuity have evolved toward more complex approaches, for example conditioned on target classes. Artifactual issues accompanying real-world bioassay data have also been an important consideration, that is, false positives associated with assay methodologies. Thus making effective use of bioassay data requires understanding promiscuity both as a biological phenomena and experimental source of error. Mindful of these new challenges, we have developed a component for BARD (BioAssay Research Database) called the 'Badapple Promiscuity Plugin' (BioActivity Data Associative Promiscuity Pattern Learning Engine). The Badapple algorithm generates a score, based on scaffold-family membership, and derived solely from empirical activity data in BARD. The score reflects both a pan-assay 'batting average' and the weight of evidence, thus high scores indicate confidence. 'Evidence-based' or 'data-driven' further implies that the algorithm evaluates data 'as is', thus scores may change as new evidence becomes available. Fully integrated with BARD via the flexible IPlugin specification, BARD's semantic advances provide a valuable and unique synergy. The new annotations and bioassay ontology, based on BAO and adapted and implemented for BARD, enable improvements, extensions, and customizations for Badapple. In this presentation we will describe these tools and some examples from molecular discovery scenarios.

BADAPPLE promiscuity plugin for BARD: Evidence-based promiscuity scores

246th National Meeting (2013)

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