Evidence-Based and Quantitative Prioritization of Tool Compounds in Phenotypic Drug Discovery

The use of potent and selective chemical tools with well-defined targets can help elucidate biological processes driving phenotypes in phenotypic screens. However, identification of selective compounds en masse to create targeted screening sets is non-trivial. A systematic approach is needed to prioritize probes, which prevents the repeated use of published but unselective compounds. Here we performed a meta-analysis of integrated large-scale, heterogeneous bioactivity data to create an evidence-based, quantitative metric to systematically rank tool compounds for targets. Our tool score (TS) was then tested on hundreds of compounds by assessing their activity profiles in a panel of 41 cell-based pathway assays. We demonstrate that high-TS tools show more reliably selective phenotypic profiles than lower-TS compounds. Additionally we highlight frequently tested compounds that are non-selective tools and distinguish target family polypharmacology from cross-family promiscuity. TS can therefore be used to prioritize compounds from heterogeneous databases for phenotypic screening.