SpatioCAD: Context-aware graph diffusion model for pinpointing spatially variable genes in heterogeneous tissues.

SpatioCAD, a cell-density-aware graph diffusion model, identifies spatially variable genes in heterogeneous tumor tissues with improved statistical power and specificity across simulated datasets.

• Why it matters: Existing SVG methods often ignore cell density variation, which can create false positives in complex tissues like tumors.
• What they did: SpatioCAD decouples true spatial expression from cellularity-driven confounding by simulating expression propagation on a graph with cell-density-aware conditions.
• The result: In breast cancer, lung cancer, and glioma spatial transcriptomics datasets, it recovers biologically meaningful tissue architecture and functionally diverse SVGs, including low-abundance transcripts, while remaining fast and scalable.

AI summary of the abstract · may contain errors · not medical advice · read the paper.