Intuitive intracellular communication deconvolution and ranking for improved discovery of context-de
Updated: Sep 29
Michael E. Troka, School of Dental Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA. Michael V. Gonzalez, Center for Cytokine Storm Treatment & Laboratory, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA. Dana T. Graves, Department of Periodontics, School of Dental Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
Variations in cell-cell communication between cellular states, such as disease status, often define biologically relevant phenotypes observed at the organismal level. Recently, single cell transcriptomics has allowed for the exploration of cell-type heterogeneity and gene expression signatures across a wide array of tissues and cellular states. There are several existing cell-cell communication tools designed to decipher ligand-receptor interactions, however, these computational tools lack features and easily interpretable outputs which make discovery, validation, and systematic ranking of signaling pairs driving biologically significant and distinct states difficult. Some shortcomings of current cell-cell communication packages include: a) convoluted ranking metrics; b) inability to distinguish biologically relevant ligand-receptor variability from scRNAseq protocol-related technical noise; c) limited flexibility for multiple condition and variable comparison; d) use in confirmatory rather than discovery-based experiments; e) native database limitations to either mouse or human samples; and f) minimal features for interpretation and downstream analysis. To address these shortcomings, we introduce TrokaChat, a computational tool for improved discovery and ranking of biologically relevant ligand-receptor pairs in sparse scRNAseq data. TrokaChat allows for versatile comparison between biological states with powerful downstream analysis capabilities to direct further signaling-derived hypothesis testing. TrokaChat produces a communication score between ligand-receptor pairs by incorporating adjusted log fold change and cluster specific gene expression values. Furthermore, we apply TrokaChat and other existing cell-cell communication tools to publicly available scRNAseq data from atopic dermatitis patients. TrokaChat regularly outperformed other cell-cell communication tools by producing higher relative ranking for experimentally verified, biologically relevant ligand-receptor interactions. Designed to integrate with existing Seurat-based workflows, TrokaChat is open source and incorporates condition-wise ligand-receptor pair GSEA, informative visualizations, multi-condition comparison, and an adapter for further tensor-based analysis. Therefore, TrokaChat allows for the intuitive discovery of biologically relevant cell signaling across conditions through an easy to interpret ligand-receptor ranking system.