Inference of cell-cell communication from single-cell RNA sequencing data is a powerful technique to uncover intercellular communication pathways, yet existing methods perform this analysis at the level of the cell type or cluster, discarding single-cell-level information. Researchers at the Stanford University School of Medicine have developed Scriabin, a flexible and scalable framework for comparative analysis of cell-cell communication at single-cell resolution that is performed without cell aggregation or downsampling. The researchers use multiple published atlas-scale datasets, genetic perturbation screens and direct experimental validation to show that Scriabin accurately recovers expected cell-cell communication edges and identifies communication networks that can be obscured by agglomerative methods. Additionally, they use spatial transcriptomic data to show that Scriabin can uncover spatial features of interaction from dissociated data alone. Finally, the researchers demonstrate applications to longitudinal datasets to follow communication pathways operating between timepoints. This approach represents a broadly applicable strategy to reveal the full structure of niche-phenotype relationships in health and disease.
Schematic overview of cell-resolved communication analysis with Scriabin
Scriabin consists of multiple analysis workflows depending on dataset size and the user’s analysis goals. a, At the center of these workflows is the calculation of the CCIM M, which represents all ligand–receptor expression scores for each pair of cells. b, CCIM workflow. In small datasets, M can be calculated directly, active CCC edges predicted using NicheNet and the weighted cell–cell interaction matrix used for downstream analysis tasks, such as dimensionality reduction. M is a matrix of N × N cells by P ligand–receptor pairs, where each unique cognate ligand–receptor combination constitutes a unique P. c, Summarized interaction graph workflow. In large comparative analyses, a summarized interaction graph S can be calculated in lieu of a full dataset M. After high-resolution dataset alignment through binning, the most highly variable bins in total communicative potential can be used to construct an intelligently subsetted M. d, Interaction program (IP) discovery workflow. IPs of co-expressed ligand–receptor pairs can be discovered through iterative approximation of the ligand–receptor pair TOM. Single cells can be scored for the expression of each IP, followed by differential expression and modularity analyses.
Availability – Scriabin is available for download and use as an R package at https://github.com/BlishLab/scriabin