Fig. 5

Cell neighborhood and cell-to-cell communication analyses. a Schematic of a multicellular tissue with an organized cellular structure (left) and environment specific gene expression (right). b A network representation of the pairwise interacting cell types identified by Giotto in the seqFISH+ somatosensory cortex dataset. Enriched or depleted interactions are depicted in red and green, respectively. Width of the edges indicates the strength of enrichment or depletion. c Visualization of the cell-to-cell communication analysis strategy. For each ligand-receptor pair from a known database a combined co-expression score was calculated for all cells of two interacting cell types (e.g., yellow and blue cells, left). This co-expression score was compared with a background distribution of co-expression scores based on spatial permutations (n = 1000). A cell-cell communication score based on adjusted p value and log2 fold change was used to rank a ligand-receptor pair across all identified cells of interacting cell types (right). d Heatmap (left) showing the ranking results for the ligand-receptor analysis as in c (y-axis) versus the same analysis but without spatial information (x-axis) for all the ligand-receptor pairs. AUC plot (right) indicating the percentage of expression ranks that need to be considered to recover all the first spatial ranks. e Dotplot for ligand-receptor pairs that exhibit differential cell-cell communication scores due to spatial cell-cell interactions. The size of the dot is correlated with the adjusted p value and the color indicates increased (red) or decreased (blue) activity. Dots highlighted with a green box are used as examples in f. f Heatmaps showing the increased expression of indicated ligand-receptor pairs between cells of two interacting cell types. g Barplot showing gene expression changes in subsets of endothelial cells (left) stratified based on their spatial interaction with other indicated cell types (right, schematic visualization)