Genetically informed single-cell and spatial mapping of metabolic programs in human health and disease

Defining cell-type-specific endogenous metabolic features, the spatial distribution of cell-level metabolic states and cellular responses to exogenous metabolites is very important for understanding disease mechanisms. However, existing transcriptome-based metabolic models primarily infer intracellular reaction or pathway-level activities, and therefore cannot directly assess associations between individual metabolite levels and cellular states, particularly for metabolites that act extracellularly as signalling molecules rather than entering cells as metabolic substrates. To overcome this problem, we introduce the gmMAP (Genetically informed metabolite trait mapping across single-cell and spatial tissues), a framework that integrates metabolite GWAS summary statistics with single-cell and spatial transcriptomes to map metabolic programmes at cellular and spatial resolution. Notably, the gmMAP enables the prediction of endogenous metabolic process activation while also revealing intrinsic associations between exogenous metabolites and diverse cellular functional states. To further capture the connectivity of cellular metabolic networks, we incorporated a constraint-based metabolic flux model to evaluate global metabolic activity. To evaluate the accuracy and generalizability of gmMAP, we applied the framework across representative biological contexts spanning human development, physiological homeostasis, inflammation and cancer. In human kidney development, the gmMAP captured dynamic metabolic programmes, which was validated using paired transcriptomic and metabolomic reference datasets, supporting its reliability in metabolite identification and metabolic-flow inference. At the organ level, the gmMAP reconstructed spatial metabolite distribution patterns across 17 mouse organs under homeostatic and autoimmune inflammatory conditions, and further extension of gmMAP to 24 normal human tissues generated a multi-scale metabolic atlas at both organ and cellular resolutions. In disease settings, gmMAP revealed metabolic reprogramming across 29 pan-cancer cell populations, and identified potential links between exogenous metabolites and inflammation-associated stromal metabolic remodelling in ulcerative colitis. Together, gmMAP can consistently connect genetically informed metabolite traits with cell states, spatial tissue organization and disease pathology.