Extended Data Fig. 2: Enrichment of subtype-specific gene programs (GP) and neoplastic cell phenotypes in chemo-naïve and post-CTX samples.
From: Persister cell phenotypes contribute to poor patient outcomes after neoadjuvant chemotherapy in PDAC
a) Heatmap and bar plot of previously defined Bailey GP signatures that are significantly enriched in either the Classical/Progenitor/Immunogenic, Basal-like/Squamous or ADEX subtypes. Heatmap z scores represent changes in median GP scores between chemo-naïve (n = 64) and post-CTX (n = 33) samples. Bar plots represent –Log10 (Kruskal Wallis rank sum test two-sided P-value adjusted for multiple testing (pAdj)). b) Single sample gene set enrichment analysis using GP signatures (as shown in a) followed by tSNE. Sample clusters are identical between tSNE plots with individual samples highlighted according to treatment type, chemotherapy regimen and indicated GP score. Analyses for chemo-naïve (n = 64) and post-CTX (n = 33) samples are shown. c) Bar charts showing the differential enrichment of specific neoplastic cell populations in chemo-naïve and post-CTX samples as indicated. The significance of the enrichment is provided as -Log10(Wilcoxon rank sum test two-sided pAdj-value adjusted for multiple testing). The dotted line represents -Log10(pAdj ≤ 0.05). d) Volcano plots showing the differential enrichment of genes associated with the indicated neoplastic cell populations in post-CTX samples treated with either GEM or mFFX. Genes significantly enriched (Log 2 Fold change > 1 and -Log10(pAdj) >2) in samples treated pre-operatively with GEM or mFFX are shown. pAdj represent the significance of a two-sided Wald test adjusted for multiple testing.
