Extended Data Fig. 6: H2A.Z occupancy associates with chromatin opening and DNA methylation.

a, Chromatin accessibility as a function of H2A.Z occupancy. Chromatin accessibility log₂FC (y-axis) was measured by DESeq2-normalized ATAC–seq Tn5 insertion counts in MED12 (n = 4), HMGA2 (n = 4), FH (n = 4) and YEATS4 (n = 4) tumours compared with normal samples (n = 15) at H2A.Z peaks stratified into increased, no change and decreased binding. b, c, Pileup of H2A.Z ChIP fragments from pooled myometrium data at DARs shown by composite plots and heatmaps. DARs in each UL subclass (b) or in YEATS4 tumours stratified by overlap with other UL subclasses (c) are represented by heatmap rows over which mean H2A.Z fragment coverage is calculated. d, Differences in H2A.Z binding at DARs in UL subclasses as compared to normal samples. Violet dots represent differential H2A.Z binding sites (FDR < 0.05, |log₂FC| > 1). The highlighted peaks (pink dots) are located close to CBX8 and named as S with distance in kilobases from the CBX8 TSS. e, Mean sample-wise DNA methylation differences in MED12 (n = 11), HMGA2 (n = 26), FH (n = 6), YEATS4 (n = 14) and OM (n = 8) tumours compared with respective normal samples at H2A.Z peaks stratified into increased, no change and decreased binding. H2A.Z binding differences in a, d, e are from the spike-in ChIP–seq experiments comparing two tumours from each subclass to four normal samples. Increased, no change and decreased binding were defined using FDR < 0.05 and |log₂FC| > 1 cutoffs. Boxplots show the median and the first and third quartiles. Error bars extend up to 1.5 IQR beyond the quartiles.