Extended Data Fig. 8: The pro-angiogenic role of stromal cell subtypes, related to Fig. 5.
a, DEG expression across mesenchymal subtypes. b, Heatmap showing distinct disease prevalence estimated by Ro/e. The top barplot shows cell composition and the right barplot shows sample composition across distinct diseases. c-e, Boxplots compare the proportions of mesenchymal subtypes between Non-tumor (n = 16) and Adeno_Pri (n = 75) (c), paired NAT (n = 21) and Adeno_Pri (n = 21) (d), and P (n = 20) and P_Mets (n = 55) (e). f, Boxplots showing tissue site preference of F_C7_COL4A2. g-h, The DEGs (g) and enriched pathways (h) of F_C1_CFD_iCAF between P and P_Mets. i, VEGFB expression across mesenchymal subtypes. j, Kaplan-Meier plots illustrating OS of patients grouped by the percentage of F_C1_CFD_iCAF. k-l, The DEGs (k) and enriched pathways (l) of endothelial cell subtypes. m, Heatmap showing distinct disease prevalence estimated by Ro/e. The top barplot shows cell composition and the right barplot shows sample composition across distinct diseases. n-p, Boxplots compare the proportions of endothelial cell subtypes between Non-tumor (n = 16) and Adeno_Pri (n = 75) (n), paired NAT (n = 21) and Adeno_Pri (n = 21) (o), and P (n = 20) and P_Mets (n = 55) (p). q, Pearson correlation between the percentages of EC_C2_CXCR4_Tip and EC_C0_ACKR1_Vein (left panel) or EC_C3_GJA5_Artery (right panel), with 95% confidence interval shown. The two-sided t-test was used. r, FLT1 expression across classic endothelial cell subtypes. s, Kaplan-Meier plots illustrating OS of patients grouped by the percentage of EC_C3_GJA5_Artery. For c, d, e, n, o, and p, the two-sided Wilcoxon test was used. For f and g, the Benjamini-Hochberg-adjusted two-sided Wilcoxon test was used. For h and l, the Benjamini-Hochberg-adjusted one-sided hypergeometric test was used. For j and s, the log-rank test was used. For boxplots in c, d, e, n, o, and p, the center line represents the median, box edges show quartiles, whiskers extend 1.5 times the interquartile range, and points beyond are individual outliers. n represents biologically independent samples.
