Interpretable niche-based cell‒cell communication inference using multi-view graph neural networks

Dr. Chuanyun Li published a paper in Nature Computational Science with his collaborators.

Cell‒cell communication (CCC) is a fundamental biological process for the harmonious functioning of biological systems. Increasing evidence indicates that cells of the same type or cluster may exhibit different interaction patterns under varying niches, yet most prevailing methods perform CCC inference at the cell type or cluster level while disregarding niche heterogeneity. Here we introduce the Spatial Transcriptomics-based cell‒cell Communication And Subtype Exploration (STCase) tool, which can describe CCC events at the single-cell/spot level based on spatial transcriptomics (ST). STCase includes an interpretable multi-view graph neural network via CCC-aware attention to identify niches for each cell type and uncover niche-specific CCC events. We show that STCase outperforms state-of-the-art approaches and accurately captures reported immune-related CCC events in human bronchial glands. We also identify three distinct niches of oral squamous cell carcinoma that may be obscured by agglomerative methods, and discover niche-specific CCC events that could influence tumor prognosis.