Fig. 5

hMeDIP-seq hydroxymethylation profiling in the brain. a Scatter plot showing the correlation between hMeDIP-seq replicates. For 300-bp genomic tiles logCPM (count per million) values were calculated for each replicate. Each dot represents one genomic tile. b Bar plot showing observed over expected by chance enrichment of 5hmC peaks detected by hMeDIP-seq at multiple genomic locations; computationally derived chromatin segmentation (ChromHMM) of Brain Frontal Lobe genome, as well as CpG islands and CpG island shores. c, d Correlation of 5hmC profiling between WG Bis/OxBis and hMeDIP-seq. c The distribution of WG-derived 5hmC levels at hMeDIP summits expanded ± 150 bp and gaps between them (observed, left). Random permutation of hMeDIP summits and gaps and 5hmC levels distribution (expected, right). The differences between gaps and summits are statistically significant as determined by Kruskal–Wallis nonparametric test (p < 0.01). d The mean of the differences of 5hmC levels between hMeDIP-seq summits and gaps (observed) and the distribution of the mean for the permuted summits and gaps (expected). e Schematic representation of hMeDIP-seq peaks with no 5hmC detected by WG Bis/OxBis-seq (average 5hmC = 0) and with 5hmC detected by WG Bis/OxBis-seq (average 5hmC > 0). f Pie charts showing the proportion of real hMeDIP peaks (top) and randomly permuted hMeDIP peaks (bottom) with and without 5hmC detected by WG Bis/OxBis-seq. The distribution of the proportion of randomly permuted hMeDIP peaks with 5hmC (expected) and the actual proportion of hMeDIP peaks is shown on the right. For the analysis, hMeDIP-seq peaks with all associated CpGs having at least 10× WG coverage have been selected, which accounts for approximately 9000 peaks. g Genomic regions showing specific (left) and non-specific (right) hMeDIP-seq peaks