Objective

Several studies have shown that biosignatures derived from microscopy images and high-dimensional transcriptional readouts such as L1000 contain information about target activities and mechanism-of-action (MoA) of small molecules. In this talk we will discuss applications of such biosignatures in drug discovery. (1) In the context of an antiviral screen, we profiled the obtained high-throughput screening (HTS) hits via a cell painting-like imaging assay to characterize the different compound MoAs. This enabled us to distinguish molecules acting on host-cell targets from direct antivirals, resulting in fast identification of hit-to-lead candidates. (2) For HTS and hit expansion purposes, we applied biosignatures to rank and select compounds to be tested in the respective screening assay. By re-using existing historical screening data, including those from large-scale imaging-based screens, we were able to predict the biological activities of compounds in other assays. This resulted in increased hit rates and identification of additional chemical scaffolds compared to those obtained in the original screen. Taken together, biosignatures can increase the efficiency of screening campaigns and reduce the need for costly and time-consuming screening cascades with multiple profiling assays.