Skip to main content
Fig. 3 | Epigenetics & Chromatin

Fig. 3

From: Modelling the conditional regulatory activity of methylated and bivalent promoters

Fig. 3

Evaluation of gene-level DNA methylation scores (SMFS, MMFS, MMFR and SMRR). (Left) MMFS exhibits the strongest overall anti-correlation with RPKM-normalised transcript abundance (Pearson’s r = −0.39), indicating that it is most appropriate for capturing the gene silencing effect of promoter-localised methylation. Model performance is colour-coded by correlation within each cell line, with the best/worst-performing models highlighted in green/red, respectively. (Right) promoter methylation (MMFS) versus transcript abundance genome-wide for GM12878 (regression line shown in red), demonstrating two distinct gene clusters: active/unmethylated (green) and silent/methylated (red). It is also evident that a large number of genes exhibit near-zero expression despite a lack of substantial DNA methylation (blue); these genes reduce the predictive power of DNA methylation genome-wide and are likely silenced by other mechanisms (e.g. repressor/silencer transcription factors [22] or H3K27me3-associated Polycomb activity [16])

Back to article page