Fig. 3

Transcriptomic profiling of normal and tumor PDCOs. A Sample-to-sample distance heatmap of the RNA-seq data. The vst-transformed counts matrix was used to calculate the Euclidean distance between samples; hierarchical clustering was performed using the complete agglomeration method. Samples are grouped in two main clusters, representing normal PDCOs (top left) and tumor PDCOs derived from patient P12 (−W medium; bottom right). B Barplot showing the differentially expressed genes (DEGs; abs(log2FC) ≥ 1, FDR < 0.05) identified by comparing tumor PDCOs derived from patient P12 (−W medium) versus normal PDCOs. Early PDCOs DEGs: n = 5657 (3224 up-regulated and 2433 down-regulated); thaw PDCOs DEGs: n = 4613 (2833 up- and 1780 down-regulated); late PDCOs DEGs: n = 5290 (2746 up- and 2544 down-regulated). C Barplot of DEGs (abs(log2FC) ≥ 1, FDR < 0.05) identified applying the likelihood ratio test (LRT) to evaluate the combined effect of condition and time progression on tumor PDCOs derived from patient P12 (−W medium) versus normal PDCOs. Thaw/early PDCOs DEGs: n = 2930 (1562 up-regulated and 1368 down-regulated); late/thaw PDCOs DEGs: n = 1823 (619 up- and 1204 down-regulated); late/early PDCOs DEGs: n = 3171 (1483 up- and 1688 down-regulated). D Heatmap showing DEGs applying the LRT to evaluate the combined effect of condition and time progression on normal and tumor PDCOs gene expression. Globally, 3171 transcripts were differentially expressed in tumor PDCOs at the late time point (1483 up-regulated and 1688 down-regulated; abs(log2FC) ≥ 1, FDR < 0.05). Hierarchical clustering of transcripts and samples using the Euclidean distance and the complete agglomeration method; expression data was vst-transformed, scaled and centered. E, F The top150 significantly up- (E) and down-regulated (F) coding genes identified in the LRT analysis were annotated using the Cytoscape plugin ClueGO. Functionally enriched terms (Benjamini-Hochberg adjusted p ≤ 0.05) were identified querying the CLINVAR_Human-diseases, WikiPathways, KEGG, REACTOME_Reactions, REACTOME_Pathways, GO_ImmuneSystemProcess, GO_BiologicalProcess and CORUM_CORUM-FunCat-MIPS databases. Pie chart colors match the enriched functional clusters; the most significant terms were used as cluster representatives and identifiers