Skip to main content
Fig. 3 | Genome Biology

Fig. 3

From: CpG island composition differences are a source of gene expression noise indicative of promoter responsiveness

Fig. 3

Promoter chromatin modifications influence expression noise in mESCs. a Robust linear regression model results, modelling the influence of promoter chromatin landscape and CpG island size on expression noise. Points below the horizontal dotted line are correlated with lower noise (purple), whilst those above are associated with greater noise (orange). b Promoter chromatin modifications (fold enrichment for ChIP/input) for repressive (H3K27me3; top panel) and active (H3K4me3; bottom panel) expression are negatively correlated with expression noise. c Bivalent promoters are the noisiest promoters. Robust linear regression model results of expression noise, jointly fitting promoter category relative to the repressed promoters, with CpG island size. d The relationship between promoter chromatin modification category and rCV2. Overlaid box plots represent the median rCV2 for each category, with lower and upper quartiles. Individual points lying outside the interquartile range are plotted as black points. The y-axis is truncated at [0, 15] for clarity. mESC mouse embryonic stem cell

Back to article page