Atlas-scale metabolic activities inferred from single-cell and spatial transcriptomics

Abstract

Metabolism supplies energy, building blocks, and signaling molecules vital for cell function and communication, but methods to directly measure it at single-cell and/or spatial resolutions remain technically challenging and inaccessible for most researchers. Single-cell and spatial transcriptomics offer high-throughput data alternatives with a rich ecosystem of computational tools. Here, we present scCellFie, a computational framework to infer metabolic activities from human and mouse transcriptomic data at single-cell and spatial resolution. Applied to [~]30 million cell profiles, we generated a comprehensive metabolic atlas across human organs, identifying organ- and cell-type-specific activities. In the endometrium, scCellFie reveals metabolic programs contributing to healthy tissue remodeling during the menstrual cycle, with temporal patterns replicated in data from in vitro cultures. We also uncover disease-associated metabolic alterations in endometriosis and endometrial carcinoma, linked to proinflammatory macrophages, and metabolite-mediated epithelial cell communication, respectively. Ultimately, scCellFie provides a scalable toolbox for extracting interpretable metabolic functionalities from transcriptomic data.