#40 · Metabolomics

Inferring tissue and tumor states from spatial, single-cell, and metabolic data

The dominant thread is computational inference of tissue and cellular state from spatial or multimodal measurements. A pan-cancer spatial transcriptomics atlas across 12 cancers characterizes tertiary lymphoid structures, and H&E-based AI predicts their maturation. In PDAC, representation learning maps malignant states onto a normal foregut and pancreatic epithelial reference, and the preprint reports that developmental positioning does not explain their diversification. A separate PDAC study integrates three algorithms to infer metabolic fluxes from scRNA-seq, building a TME metabolic atlas across 460 samples and defining three subtypes. What is new is the extension of inference from spatial organization toward metabolic flux and reference-anchored state mapping. The set is heterogeneous, so this thread is dominant rather than shared by every item.