Fig. 5: Gene programs changes in active EoE and normalization in remission.

a, b Gene changes in cell-intrinsic programs in EoE. a Genes (dots) with significant differential expression (log₂(fold changes), x- and y-axis) between different conditions (axis labels). Pearson correlation coefficients and linear regression lines shown in upper left corners. The p-values are based on two-tailed one-sample Student’s t tests. b Enrichment (-log₁₀(FDR), one-tailed hypergeometric test) of Gene Ontology biological process terms (rows) in genes differentially expressed (numbers) between active EoE and both health and remission samples, in each cell type (columns). c–e Shift to IL13RA2⁺ inflammatory fibroblasts in active EoE. c Distribution of expression (y-axis, log₁₀(TP10K + 1)) in fibroblasts from the three conditions (x-axis), of genes differentially expressed in fibroblasts in EoE vs. healthy and annotated as ‘positive regulation of immune system process’. Width: Gaussian kernel density estimation of data with default parameters, scaled to have a maximum of 1; White horizontal segment: median. ***FDR < 0.001, two-tailed likelihood-ratio tests of logistic regression coefficients. d The fraction of IL13RA2⁺ fibroblasts (y-axis) and PC1 score of the centered log-ratio transformed cellular composition profile (as in Fig. 2d) for each sample (dot) from healthy (left), remission (middle) or active EoE (right) patients. The FDRs are based on two-tailed one-sample Student’s t test. e Left: Lasso linear regression coefficient (y-axis, left) of each cell type (dot) (ordered by cell type index (x-axis), as in Fig. 1d) as predictors of fraction of IL13RA2⁺ fibroblasts out of all fibroblasts. Right: Plots of lasso linear regression predictions (y-axis, based on IgG⁺ plasma cell ratios) and observed IL13RA2⁺ fibroblasts ratios. The black line is a fitted linear regression line. The p-values are based on two-tailed one-sample Student’s t test. Source data are provided as a Source Data file.