ClusToRa: A niche-centric framework for identifying structural recruitment and infiltration in spatial omics

Abstract

Spatial omics maps cellular landscapes, yet current tools might conflate stochastic proximity with organized niches. We present ClusToRa (Cluster-to-Randomization), a framework that identifies high-density cellular territories and quantifies cell-type recruitment using a fixed-position null model. Benchmarked against graph-based neighborhood-enrichment and point-pattern statistics, ClusToRa reduced false-positive enrichment in simulations and resolved core-vs-boundary interactions. Applied to cirrhotic MASH liver, ClusToRa identifies stellate-cell territories with immune/endothelial infiltration and stress-, Notch-, and PPAR-associated programs, providing a niche-centric framework for distinguishing structural cellular infiltration from boundary adjacency or density-driven colocalization.