Cell-cell communication helps to coordinate tissue development, homeostasis, and immune responses, but identifying signaling interactions within intact tissues remains difficult. Although single-cell transcriptomics has enabled systematic inference of ligand-receptor interactions, dissociation disrupts spatial context and limits the identification of bona fide local signaling and region-specific communication programs. Spatial transcriptomics and spatial multi-omics offer the opportunity to study communication in situ, but current approaches often either incompletely integrate heterogeneous modalities or return long lists of ligand-receptor pairs that are difficult to interpret mechanistically. Here, we present CellSTIC, a framework for resolving cell-cell communication in spatial multi-omics as structured communication programs grounded in tissue architecture. Rather than treating ligand-receptor interactions as isolated candidates, CellSTIC integrates multimodal evidence from local tissue neighborhoods and organizes communication into a hierarchical semantic representation that remains traceable to the underlying molecular interactions. This design enables communication to be analyzed not only at the level of individual ligand-receptor pairs but also across broader functional modules that can be compared across tissues, regions, and biological states. Across multimodal simulations with ground-truth annotations and multiple real tissue datasets, CellSTIC robustly recovered spatially coherent communication structure and spatial domains. It further resolved communication programs across three complementary dimensions: hierarchical organization in immune microenvironments, spatially restricted signaling in complex brain architecture, and developmental and regenerative remodeling across embryonic and post-injury contexts. Together, these results establish CellSTIC as a general framework for linking tissue architecture to intercellular signaling programs in situ and for generating mechanistic hypotheses from spatial multi-omic data.
Wang, S., Wang, J., Wang, J., Yuan, Z., Xu, Y.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 6
- Comments 0