Fig. 3: The landscape of CD8⁺ T cells.

a A UMAP plot illustrating six CD8⁺ T cell clusters across all samples, color-coded by cell type. b The bar plot illustrates the relative proportion of CD8⁺ T cell clusters in each sample from the Fibrosis+ LM and Fibrosis− LM. c The lollipop chart displays the prevalence of different CD8⁺ T cell subgroups between the two groups, estimated by Ro/e. The subgroups favoring the Fibrosis+ LM are positioned above, whereas those favoring the Fibrosis− LM are positioned below. d The beeswarm and box plots depict the distribution of log₂-fold differences in neighborhoods across various cell type clusters of CD8⁺ T cells, estimated by MiloR. The rightward positions indicate higher abundance in the Fibrosis+ LM, whereas the leftward positions indicate higher abundance in the Fibrosis− LM. e A heat map depicting the expression of marker genes across defined CD8⁺ T cell clusters. The color intensity reflects the average scaled gene expression. f A heat map via irGSEA visualizing the distribution of curated significant gene sets identified by robust rank aggregation across CD8⁺ T cells clusters. The number of asterisks in the grid’s upper half indicates the P value. g A Slingshot trajectory analysis of CD8⁺ T cell differentiation reveals two principal divergent trajectories. The cells are color-coded according to their pseudotime. h The density changes of CD8⁺ T cell subclusters along path 1 and path 2 are shown. i The two-dimensional plots showing the expression scores for four representative gene signatures in cells of paths 1 and path 2, respectively, along the inferred pseudotime. j The mIHC/IF images showing the positive expression of CXCL13_CD8_T_ex. k The quantification of the density of CXCL13_CD8_T_ex between Fibrosis+ LM and Fibrosis− LM, Wilcox rank-sum test.