BD2K product for enhancing phenotypic screens

Genecentrix, Inc., New York NY

Investigators

Linked publications & trials

Abstract

Project Summary A strong consensus has emerged over the past 10 years in the pharmaceutical industry that the drug-target- focused drug discovery approach, to which the industry has subscribed almost entirely since the 1980s, has been a failure. Accordingly, over the past few years, the majority of R&D in the drug discovery and development space has been re-allocated to support alternative approaches, notably phenotypic screening to discover new drugs. This shift represents a new and growing business opportunity for new, big data science and technology products, such as those developed by the applicant organization, GeneCentrix. In common phenotypic screens, a chemical library is first screened against cells (e.g. HeLa cells), and a search is then undertaken for the primary molecular target of the most active compound. This search, known as target deconvolution, is challenging and is a rate-limiting step in the success of the screen. Accordingly, new technologies have been developed for target deconvolution. Specifically, target-specific arrays of drug-like chemical probes are increasingly being used in phenotypic screens (e.g. Novartis? ?MOA box?, the NIH?s MIPE platform, etc.). GeneCentrix is pioneering the development of an information product that can enhance phenotypic screens that use these target-specific arrays. Specifically, the off-targets of the target-specific compounds (polypharmacology) in the array are currently not taken into account in analyzing the results of the phenotypic screen. In addition, the expression level of drug targets in the cell line used in the phenotype screen is very likely a predictive variable in target deconvolution but is also not taken into account. Therefore, we propose two aims to adapt our technology for the purpose of re- ranking phenotype screen results: 1) generation of off-target annotation (polypharmacologic profiles) for all compounds in a standard target-specific array; 2) integration of the polypharmacologic profile with expression data of the targets, and re-ranking of targets by this combined score. We will then validate the performance of our technology versus standard analysis of screening results; for a test set of phenotypic screens, an improvement of 20% in sensitivity associated with an equal or greater accuracy, as measured by area under the receiver operating curve (AUC), will be considered sufficient validation to pursue technical and commercial feasibility of the product in a Phase II application.