Fig. 2: Altered chromatin accessibility in recurrent tumors modulates network programming.

A Two-tailed Pearson’s correlation analysis between peak accessibility (ATAC-seq, n = 187,779 peaks) and DEGs (RNA-seq, n = 36601). Log₂FC in primary vs recurrent tumors is plotted. P-value represents a one-sample proportion test. B Chromatin accessibility matrix with gene expression markers (n = 32) by cell state (n = 20). Color indicates gene accessibility, and dot size refers to the percent of cells in the cluster. C Volcano plot depicts differential peak accessibility (Log₂FC) in primary vs recurrent tumor cells. The statistical significance of altered peak accessibility is reported as -Log10(Padj). Padj corrected for multiple comparisons. Significant peaks (n = 20832). D Functional enrichment of gained peaks as determined by GREAT v4.0.4. Metabolic and epigenetic associations are highlighted in pink. -Log10(Binomial FDR Q) is reported. Gained peaks (n = 598). E Functional enrichment of decreased peaks. -Log10(Binomial FDR Q) is reported. Decreased peaks (n = 1087). F Peak accessibility profiles by cell state with proximal gene linkages for genes (NFE2L1, NRL, MYC, TCF4, and HLX) representative of the changing cell states in primary (n = 3) and recurrent tumor (n = 3). The height and color intensity of linkages indicate the accessibility score. GABA (GABAergic), GABA-I (GABAergic-Interneuron), Glut-I (Glutamatergic-Interneuron), Glut (Glutamatergic), PN (Purkinje), CNP (Cycling Neural Progenitor), P-Photo (Progenitor-photoreceptor), P-UBC (Progenitor-UBC), RPE (Retinal Pigment Epithelium). G Top ten gene regulatory networks in primary (n = 3) compared to recurrence (n = 3). Significance of the gene network association is reported as -Log₁₀(Adj pVal). Padj corrected for multiple comparisons. H Dot plot of the top five active gene regulatory networks by cell state (n = 12). p < 0.01. See also Supplementary Fig. 3.