Identifying condition-related cell-cell communication events using supervised tensor analysis.

Abstract

Many tools have been developed to infer active cell-cell communication (CCC) events, which are essential for understanding biological processes and diseases. However, existing methods for assessing the relationships between CCC events and biological conditions have at least one practical limitation: a lack of clear interpretation, an inability to adjust for confounders, or an inability to model inherent dependencies among CCC events. To comprehensively address these limitations, we introduce STACCato, a supervised tensor analysis tool for identifying condition-related CCC events. STACCato employs a tensor-based regression model to enable statistical inference of the relationships between biological conditions (e.g., disease status or tissue types) and individual CCC events while accounting for confounders and dependencies among CCC events. Through extensive simulations and real-world applications on a lupus single-cell RNA sequencing (scRNA-seq) dataset and an autism single-nucleus RNA-seq (snRNA-seq) dataset, we demonstrate that STACCato consistently provides improved inference of condition-related CCC events compared to alternative methods. The STACCato tool is freely available on GitHub.