Fig. 8. H2BNTac promoter intensity predicts CBP/p300-regulated genes in mESCs.

a, ROC curve (left) and PRC (right) showing the performance of the ABC model and promoter ChIP–seq intensity of the indicated histone marks in discriminating CBP/p300-dependent and independent genes. CBP/p300-dependent (positive list) and CBP/p300-independent genes (negative list) were defined by nascent transcription analyses after acute CBP/p300 inhibition by A-485 (ref. ²¹). Genes downregulated (twofold or greater; average fold change of 30 min, 1 h and 2 h treatment; Methods) after A-485 treatment are considered CBP/p300-dependent; unaffected genes (<1.2-fold change) are considered CBP/p300-independent. Performance was evaluated using cumulative nominal ABC scores or ChIP–seq enrichment of H3K4me3, H3K4me1, H3K27ac and H2BNTac in promoter regions (within ±1 kb from the TSS). The ABC score was calculated using ATAC–seq, H3K27ac and Hi-C contact frequency as reported by Fulco et al.⁴⁴. b, H2BNTac promoter intensity outperforms the ABC model and other chromatin features in predicting CBP/p300 target genes. Left, AUPRC and AUROC for the ABC model. Right, comparative performance of the indicated chromatin marks. ABC scores, AUPRC and AUROC were determined as indicated in a. ABC scores were calculated using either Hi-C- or Micro-C-based contact frequency, as indicated. For the specified chromatin marks, AUPRC and AUROC were calculated using the ChIP signal in promoter regions, as defined in a. c, Schematic representation of the relative abundance of H3K4me3, H3K27ac and H2BNTac in the indicated genomic regions. H2BNTac and H3K4me3 positively discriminated active candidate enhancers and promoters, respectively.