Kontext: Kernel-based method bridging spatial domain identification and cell-cell communication

Kontext identifies spatial domains from spatial transcriptomics by jointly modeling cells and local environments, outperforming state-of-the-art methods in recovering reference domain annotations across multiple technologies.

• Why it matters: Spatial domains reflect coherent tissue organization shaped by local ligand-receptor interactions, but existing approaches often lack mechanistic insight into how those interactions organize domains.
• What they did: Kontext is a kernel-based framework that represents cells together with their local environments to ground domain identification in ligand-receptor interactions, and it outputs microenvironment, niche, and sender-receiver communication patterns.
• The result: In a Xenium temporal mouse heart cryoinjury dataset, it characterizes injury evolution and identifies an undescribed sustained complement-system activity in fibroblast–macrophage communication during scar maturation; in a million-cell 3D Open-ST human metastatic lymph node dataset, it defines the tumor–microenvironment boundary and reveals distinct interface-to-tumor-core communication patterns.

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