NicheScope: Identifying Multicellular Niches and Niche-Regulated Cell States in Spatial Transcriptomics

Abstract

The functional state of a cell is intrinsically linked to its local microenvironment, or cell niche, a complex milieu formed by multiple interacting cell types. Deciphering how these multicellular niches regulate cell states is fundamental to understanding tissue biology and disease mechanisms, yet remains challenging with current computational approaches. Here, we present NicheScope, a computational framework for transcriptome-wide identification and characterization of Multicellular Niches (MCNs) and their corresponding Niche-Regulated Cell States (NRCSs) from spatial transcriptomics data. NicheScope operates on the principle that a cells transcriptional state is associated with its local multicellular composition. It employs a robust statistical approach to jointly model neighborhood composition and transcriptome-wide gene expression, enabling the simultaneous discovery of MCNs, defined by specific combinations of neighboring cell types, and their associated NRCSs, characterized by distinct gene programs. We demonstrate NicheScopes power and versatility across diverse tissues and platforms, including lymph node, lung adenocarcinoma, and head and neck cancer. NicheScope reproducibly dissected established tissue structures in lymph nodes across tissue regions and platforms, uncovered clinically relevant tumor cell-associated MCNs in lung adenocarcinoma, and revealed shared and condition-specific MCNs in primary and metastatic tumors. Our results establish NicheScope as a powerful, robust, scalable, and interpretable framework for dissecting the spatial and functional organization of complex tissues, providing new insights into multicellular coordination in health and disease.