Accelerating Early Drug Discovery

Abstract

Many current drugs are only partially effective and have severe side effects. And, for many diseases we lack any drugs at all. There is a pressing need to accelerate the pace of effective drug discovery. We developed a dramatically different approach to traditional drug discovery programs. We make use of microscopy-based and machine-learning approaches for reading out information about the molecular states and behaviors of millions of individual cells that have been pushed—but not killed—by chemical perturbations. This information provides unique “profiles” for what each chemical does to living cells. Remarkably, we have been able to show that compounds that induce similar profiles have similar mechanisms of action on molecular pathways. This approach has the ability to identify compounds that induce cellular responses similar to those of known drugs but through different chemical structures or targets.? In this talk, we will provide an overview of this phenotypic profiling approach and its recent progress to increase the efficiency, scale and accuracy of phenotypic screens by maximizing their discriminatory power.? In addition, we will highlight the computational and statistical challenges of phenotypic profiling approaches. ?