Extended Data Fig. 5: Additional analyses of the scRNA-seq dataset of leprosy granulomas.

a) Bar graph depicting cell proportions per granuloma in the dataset of Ma, et al.⁴¹. Author-provided cell type annotations are used for analysis. b) Subclustering resolutions of T cells (left) and myeloid cells (right) required for comparative CCC analysis by agglomerative methods. Pink bars indicate the percentage of subclusters containing at least one cell from an LL granuloma and one cell from an RR granuloma. Blue bars indicate the percentage of subclusters containing at least one cell from all nine analyzed granulomas. c) NicheNet²⁰ was applied to predict ligand activities in myeloid cells between RR granulomas relative to LL granulomas. The bar plot depicts pearson coefficient outputs of NicheNet for this analysis. d) UMAP projections of T cells (top) and myeloid cells (bottom) colored by author-generated subcluster cell type annotation (left), granuloma type (middle), or if the cell falls into a cluster 2 perturbed bin (right; see Fig. 4f). e) We applied a binomial test to determine if cells from a cluster 2 perturbed bin were significantly enriched or depleted in any T cell or myeloid cell subcluster. The bar plot depicts the -log(p-value) of the exact binomial test. When p < 0.05, the bars are colored to indicate if perturbed cells are either enriched (red) or depleted (blue) from the cluster. The dotted line indicates the point at which p = 0.05. Calculated p-values are two-sided.