Characterization of clusters in scMultiome data by reference mapping. (A) Overview of reference mapping and TF prediction. (B) Feature plot displaying the prediction score calculated by the MapQuery function. (C) UMAP showing the proportion of each sample (CD: CD Patient 1–3; Control: Ctrl-Patient 1–3 in Table S1) and bar chart showing the proportion of disease and each sample. (D) UMAP plot showing the annotation of each cluster based on the prediction score from reference mapping. (E) Expression levels of selected RNA markers across identified clusters. The dot size represents the percentage of cells expressing the gene within each cluster, while the color intensity reflects the average expression level. (F) T-bet expression in each CD4⁺ T cell subset. Statistical significance for the comparisons was determined using RM one-way ANOVA, Dunnett’s multiple comparison test. n = 5 per group. *P < 0.05, **P < 0.01. (G) Heatmap showing the fold change in gene expression levels in HLA-DR⁺ T_RM cells relative to HLA-DR⁻ T_RM cells. (H) Feature plot of RUNX1 and RUNX3 expression. (I) TF activity inference calculated by the decoupleR package based on the DoRothEA network. Red indicates high activity, and blue indicates low activity.