Fig. 7: Cellular pathways associated with clinical outcomes and aberrant cell states in COPD.

a, Mediation analysis model. b, Sankey plot depicting cell-specific transcriptional modules mediating the effect of smoking or age on clinical traits. c, Sankey plot depicting cell proportion changes mediating the effect of smoking on clinical traits. d, Dot plot illustrating pathway enrichment of gene modules identified through marginal-effect analyses (circles) or mediation analyses (triangles) across key clinical traits in COPD. Clinical traits are represented along the panels on the x axis, where the direction of the effect estimate is multiplied by −log₁₀(P_enrichment). Clinical traits grouped into spirometry (FEV₁, FEV₁/FVC), emphysema (emph), exacerbations (wheezing, infections) and symptoms (cough, dyspnea, SGRQ symptoms). Pathways are displayed on the y axis. Pathways grouped into biologic themes represented by y-axis color labels (dark blue, inflammatory; orange, remodeling and repair; pink, aging hallmarks; light blue, other; yellow, cell migration and cell–ECM interactions). Significance assessed using a two-sided permutation test. The size of each dot indicates the degree of pathway enrichment, and colors correspond to specific cell types. All displayed pathways were significantly enriched within corresponding cell types after Benjamini–Hochberg correction (FDR < 0.05). e,f, Aberrant ligand–receptor signaling. e, Receiver-focused analysis. Rows represent ligand–receptor pairs (or ligands alone when the cognate receptor was not profiled with Xenium). Columns indicate either ligand-expressing (left) or receptor-expressing (right) cell types. Color intensity in the left reflects normalized ligand expression across sender populations measured by snRNA-seq. Right, the normalized difference in hotspot prevalence between inflamed and noninflamed neighborhoods, calculated as (hotspot % in inflamed − hotspot % in noninflamed)/(hotspot % in inflamed + hotspot % in noninflamed). For IM_CHIT1, contrasts were computed relative to IM₁. f, Sender-focused analysis. Rows represent ligand–receptor pairs (or ligands alone if the receptor was not profiled with Xenium). All ligands shown were significantly upregulated in inflamed versus noninflamed senders, IM_CHIT1 versus IM₁ or Fibro_CTHRC1 versus other fibroblasts (Wilcoxon rank-sum test, FDR < 0.05). Left, the log₂(FC) in ligand expression relative to the matched control population. Right, the normalized difference in hotspot prevalence using the same formula as in e. Only ligand–receptor interactions with significant differences in both hotspot prevalence (two-proportion z test; FDR < 0.05) and hotspot intensity (Moran’s I z score, Wilcoxon rank-sum test; FDR < 0.05) are shown.