Understanding cell–cell communication (CCC) is a cornerstone of spatial biology, providing insight into the molecular crosstalk within tissue microenvironments. The relationship between a ligand expressed in one cell state and its cognate receptor in a co-localized partner cell state underpins tissue function and organization. Recent technologies such as Xenium, MERSCOPE, CosMx, seqFISH and Slide-seqV2 have made CCC prediction feasible. However, challenges remain, including low sensitivity and limited gene counts, that hinder the clustering of spatial cells and cell-type identification. The computational integration of spatial transcriptomic data with analogous single-cell RNA-sequencing (scRNA-seq) data offers a powerful approach to address these limitations.

Second, NiCo infers interactions between cell types in spatially co-localized regions by constructing a neighbourhood matrix that captures all instances of cell-type co-localization across the tissue; NiCo uses this matrix as input to a classifier that predicts interaction strengths for every pair of cell types. Positive interaction strengths indicate likely interactions between cell types, and these predictions have been rigorously validated using simulated datasets. This capability allows researchers to infer the functional relationships between cell types and assess the spatial organization of cellular interactions.