Quantitative microscopy has proven a versatile and powerful phenotypic screening technique. Recently, image-based profiling has shown promise as a means for broadly characterizing molecules’ effects on cells in several drug-discovery applications, including target-agnostic screening and predicting a compound’s mechanism of action (MOA). Several profiling methods have been proposed, but little is known about their comparative performance, impeding the wider adoption and further development of image-based profiling. We compared these methods by applying them to a widely applicable assay of cultured cells and measuring the ability of each method to predict the MOA of a compendium of drugs. A very simple method that is based on population means performed as well as methods designed to take advantage of the measurements of individual cells. This is surprising because many treatments induced a heterogeneous phenotypic response across the cell population in each sample. Another simple method, which performs factor analysis on the cellular measurements before averaging them, provided substantial improvement and was able to predict MOA correctly for 94% of the treatments in our ground-truth set. To facilitate the ready application and future development of image-based phenotypic profiling methods, we provide our complete ground-truth and test datasets, as well as open-source implementations of the various methods in a common software framework.

The application of high-content imaging in conjunction with multivariate clustering techniques has recently shown value in the confirmation of cellular activity and further characterization of drug mode of action following pharmacologic perturbation. However, such practical examples of phenotypic profiling of drug response published to date have largely been restricted to cell lines and phenotypic response markers that are amenable to basic cellular imaging. As such, these approaches preclude the analysis of both complex heterogeneous phenotypic responses and subtle changes in cell morphology across physiologically relevant cell panels. Here, we describe the application of a cell-based assay and custom designed image analysis algorithms designed to monitor morphologic phenotypic response in detail across distinct cancer cell types. We further describe the integration of these methods with automated data analysis workflows incorporating principal component analysis, Kohonen neural networking, and kNN classification to enable rapid and robust interrogation of such data sets. We show the utility of these approaches by providing novel insight into pharmacologic response across four cancer cell types, Ovcar3, MiaPaCa2, and MCF7 cells wild-type and mutant for p53. These methods have the potential to drive the development of a new generation of novel therapeutic classes encompassing pharmacologic compositions or polypharmacology in appropriate disease context. Mol Cancer Ther; 9(6); 1913-26. ©2010 AACR.