Fig. 2

Oxidative damage distribution is associated with genomic features. a Bar plot displays the average correlation of damage levels with large-scale chromatin and other features in HepG2 cells at 100-kb resolution. Damage correlates with euchromatic features and anticorrelates with heterochromatic ones, the opposite of that observed for cancer SNVs. The heatmap shows the relationship between the features, grouped using hierarchical clustering. b The plot shows dependence between Relative Enrichment of damage and genomic GC content at 1-kb resolution. Damage levels increase with GC content and then surprisingly fall in high GC areas. The blue line marks the genomic average GC content of 41%. c Metaprofile of Relative Enrichment over ~ 23,000 protein-coding genes (n_genes = 23,056, n_promoters = 48,838, n_5UTRs = 58,073, n_exons = 214,919, n_introns = 182,010, n_3UTRs = 28,590, n_termination = 43,736, n_intergenic = 22,480). Damage levels for UTRs, exons, introns, and intergenic regions are averaged across each feature due to their variable sizes. GC content is depicted for the same regions smoothed with a Gaussian smooth ranging over 100 bp. Coding and regulatory regions are depleted for damage despite their increased GC content, whereas introns have near intergenic damage levels. d, e Boxplots depict damage levels at 48,838 promoters binned into unexpressed and expression deciles (d) and average GC content deciles (e). Promoters are defined as the transcriptional start sites ± 1 kb. Damage is not transcription-dependent but reduces with increasing promoter GC content. f, g Metaprofiles of Relative Enrichments and average GC contents across 848,350 Alu and 2533 LINE elements. There is a very large accumulation of damage inside these features. All panels display relative AP-site enrichment for X-ray-treated samples; for corresponding plots of the other treatment conditions, see Additional file 1: Figure S4A-D. Error bars and shaded borders show the standard error of mean across three biological replicates