Relationships between RNA half-lives, transcript levels and chromatin marks. (a) Boxplot chart of the RNA half-lives of mRNAs of varying expression levels. (b) Boxplot charts of the signal intensities for the ChIP-seq peaks plotted for gene populations with different RNA half-lives, as indicated on the x-axis. (c) Boxplot chart of the half-life distributions in different ChIP-seq and RNA-seq fractions. Labels on the x-axis refer to the quadrant from Figure 1b. For ChIP (+) / RNA (+), ‘total’ indicates the total transcripts, and ‘×2’ and ‘×1.1’ indicates all the genes within 2-fold (green dots in Figure 2d) and 1.1-fold (red dots in Figure 2d) of the least squared regression line, respectively. Asterisks in (a–c) indicate statistical significance by Wilcoxon’s signed rank test (P-values *P < 0.05, **P < 0.01, ***P < 0.001). (d) Scatterplot to show the distributions of ChIP (+) / RNA (+) genes. Line is the least squared regression line. Red and green dots indicate genes within 1.1-fold and 2-fold of the least squared regression line, respectively. (e) Scatterplot representing the ChIP-seq peak signal intensities of H3K4me3 on the y-axis and gene expression values on the x-axis for transcripts with ChIP-seq peaks and measured half-lives (total n = 12,479). Genes having “short” (t1/2 < 4 h) half-lives (short n = 3190) are indicated in red dots and were statistically significant (p-value 6.8 × 10−16). Dotted lines show the 10 RPKM and 1 × 104 peak intensities.