Extended Data Fig. 10: Spatial analysis of ulcer bases in Crohn’s disease.

Related to Fig. 3. A. UMAP embedding visualising all cells from integrated Xenium in situ 5100-plex analysis, cells are coloured by broad cell lineage. n = 19 samples, n = 1,860,368 cells. B. As in A, except cells are coloured by sample type. Colour legend as in D. n = 19 samples, n = 1,860,368 cells. C. Sample pseudobulk PCA analysis of all tissue sections analysed in the 5100-plex ST cohort. Colour legend as in D. n = 19 samples. D. As in C, except pseudobulk PCA was calculated only on cells from lesion (fistula or ulcer) areas as defined by niche analysis (Methods). Healthy control samples were excluded from the analysis, as no lesional areas were present. n = 12 samples. E. Representative tissue sections (n = 19 samples) visualising broad cell type tissue distribution in healthy control tissue (left), CD ulcer (middle) and fistula tract (right). Colour legend as in A. F. UMAP embedding visualising all fibroblast subpopulations detected from clustering analysis of integrated Xenium in situ 5100-plex ST cohort, coloured by detected fibroblast subclusters. N = 19 samples. G. As in F, except points are coloured by sample type. n = 19 samples. H. Dotplot visualising top marker gene expression of fibroblast subtypes visualised in F. I. A representative CD inflammatory tissue section profiled with Xenium 5100-plex ST panel (n = 19 samples). Colour legend as in A. A white box with dashed lines indicates an ROI with ulcer. J. FAS cell type distribution around the ROI indicated in panel i. All other cells are shown in grey. N = 19 samples in cohort. K. Selected gene expression of FAS and macrophage cell marker genes in the ulcer ROI indicated in panel I. n = 19 samples in cohort. L. FAS cell type distribution around two representative fistula tracts profiled with Xenium 5100-plex analysis. Colour legend as in k, all other cells are shown in grey. N = 19 samples in cohort. M. cNMF analysis of fibroblast cells from Xenium 5100-plex in situ ST cohort, with pie charts indicating average disease associated factor usage per cluster. Factor colours are as in bar charts below in panel n, with other factors shown in grey. N. Barplots visualising cNMF factor top gene loadings for selected factors enriched in disease-associated fibroblast clusters. O. Volcano plot showing differential expression analysis between ulcer vs fistula FAS-ALC cells. Differential expression was assessed using DESeq2 with Wald tests and Benjamini–Hochberg correction for multiple testing. Testing was carried out on pseudobulk expression in n = 19 samples. P. Differential abundance analysis of fibroblast subpopulations presented in panel F, comparing cells from CD ulcer samples with healthy controls (n = 19 samples in cohort). Bands represent the 95% Bayesian credible interval of the slope (logit fold change in cluster proportion per unit change in the covariate), indicating the range of effect sizes compatible with the data, given the model. Q. Differential abundance analysis of fibroblast subpopulations presented in panel F, comparing cells from CD fistula samples with healthy controls (n = 19 samples in cohort). Bands represent the 95% Bayesian credible interval of the slope (logit fold change in cluster proportion per unit change in the covariate), indicating the range of effect sizes compatible with the data, given the model. R. Differential abundance analysis of fibroblast subpopulations presented in panel F, comparing cells from CD fistula samples with CD ulcer samples (n = 19 samples in cohort). Bands represent the 95% Bayesian credible interval of the slope (logit fold change in cluster proportion per unit change in the covariate), indicating the range of effect sizes compatible with the data, given the model.