“Building an Integrative View of Immunity with Data Tensors”

High-throughput technologies now allow us to characterize immune responses with comprehensive scale and detail. While profiling cells across parameters like time, perturbations, and genetics yields rich data, these multidimensional datasets challenge conventional analysis. Standard workflows typically flatten data into two-dimensional matrices, sacrificing the essential context provided by the experimental structure. Instead, I will show how tensor decompositions can preserve the structure of multidimensional studies. Choosing the right mathematical representation allows us to derive robust insights—from utilizing in vitro characterization to optimize engineered cytokines, to identifying single-cell features that define autoimmunity. Furthermore, these approaches enable integrative analysis that links cellular states to alterations at the tissue, organ, and individual scales.