Spatial transcriptomics-metabolomics mapping with SCrOFit

Spatial transcriptomics (ST) and spatial metabolomics (SM) provide complementary insights into cellular communication within complex tissues, yet their integration is hindered by mismatched resolution and dimensionality. We introduce Spatial Cross-Omics Fitting (SCrOFit), a computational framework that enables cellular-resolution alignment of SM and ST data. SCrOFit leverages a minimum-cost optimal transport model to couple SM pixels with ST-defined cell identities, effectively bridging the resolution gap, achieving precise cross-modality integration. This approach reveals the transcriptomic mechanisms underlying metabolomic features at the single-cell level. Applying SCrOFit to the aging mouse brain, we identify a distinct spatial metabolic reprogramming in the hippocampus, characterized by a transition from glycogen catabolism to ketone body utilization. Mechanistically, we show that downregulation of phosphorylase kinase in aging astrocytes serves as a molecular bottleneck for glycogen degradation, driving these observed metabolic shifts. SCrOFit provides a robust tool for spatially resolved multi-omics analysis, offering new opportunities to dissect molecular dynamics across diverse biological contexts.