This event will be held virtually via Zoom (check email or contact ksas@seas.upenn.edu).

Human diseases are fundamentally multicellular in nature with many different cell types contributing to disease progression and treatment response. However, how therapeutics impact each cell type in a heterogeneous population remains poorly understood because most studies are focused on isolated cell types or a handful of pathways. Now, single-cell transcriptional profiling methods allow us to collect a deep molecular portrait of the collective response of heterogeneous populations of cells to any perturbation. In my talk, I will present my research in harnessing the power of single-cell transcriptional profiling measurements to dissect therapeutic response in heterogeneous cell populations. In the first part, I will describe the probabilistic modeling framework I developed for analyzing single-cell population data across perturbations at scale (PopAlign). PopAlign models single-cell data with semantically interpretable, low-error, highly-compressed probabilistic models, which allows fast comparisons across hundreds of samples. In the second part, I will discuss how I applied this framework to analyze a drug response study of over 1.6M human primary immune cells to 500 commercially-available immunomodulatory compounds. While most compounds in the library exert broad impact across multiple cell types in the population, my analysis also reveals highly cell-type specific activity, including a novel myeloid-suppressing function of a group of compounds including NSAIDs and an artificial sweetener. My work provides new depth and insight into how existing compounds reshape immune populations, and a general platform for evaluating and designing population-level responses to therapeutic interventions.