Figure 7.
Analysis of differential methylation between alleles. (A) GenPlay genome browser view illustrating an a-DMR between parental alleles on chromosome 8. Tracks 1 and 3 illustrate smoothed average methylation levels for the paternal (blue) and maternal (red) alleles for the indicated individual; tracks 2 and 4 highlight the corresponding a-DMRs identified by DSS software as described in the “Methods” section. (B) A-DMRs are richer in heterozygous SNPs than the rest of the genome. Data were generated using SnpEff based on the previously published genome sequence of FNY01_3_2 and 3_3. The expected number of SNPs was calculated by shuffling the a-DMRs 10 000 times at random location in the genome where the coverage was similar to that of the a-DMRs. The average of ∼1 SNP/kb in the expected samples is very close to the average number of SNPs/kb genome-wide observed in these 2 individuals. (C) Bar graphs illustrating the replication timing of HMDs, short PMDs, and long PMDs in 3 cell types. Average replication timing was determined for 5-kb tiles genome-wide and separated into quartiles. The fractions of each domain in the respective quartile of replication timing are indicated. In BasoE, IMR90, and K562 cells, both HMDs and short PMDs replicated early, whereas long PMDs replicated late. (D) Scatter plots illustrating the correlation between timing of replication and DNA methylation in c-ARDs. x-axis: differential allelic replication timing ratio = maternal timing ratio − paternal timing ratio, with timing ratio being the sum of maternal or paternal TimEX read counts in S phase over the sum of maternal or paternal TimEX read counts in G1 phase, respectively, across individual regions; y-axis: differential allelic methylation = βmat − βpat with β = ratio of the sum of maternal or paternal methylated reads over the sum of maternal or paternal total read counts, respectively, across individual regions. Black dots, regression line and Pearson r2 for all c-ARDs; red dots, regression line and Pearson r2 for c-ARDs with an abs(Δ DNA methylation) >0.05. There is no correlation between timing of replication and DNA methylation in c-ARDs except for a subgroup of outlier c-ARDs (in red) that overlap with a-DMRs. (E) Scatter plots illustrating the correlation between timing of replication and DNA methylation in a-DMRs. X-axis: differential allelic methylation (as in panel D) across individual a-DMRs; y-axis: differential allelic replication timing ratio (as in panel D). All a-DMRs and c-ARDs plotted contained at least 300 informative reads for allele-specific measurement of DNA methylation or for timing of replication analysis. Black dots, regression line and Pearson r2 for a-DMRs with an abs(Δ DNA methylation) >0.05; red dots, regression line and Pearson r2 for a-DMRs with an abs(Δ DNA methylation) >0.1. There is a strong correlation between timing of replication and DNA methylation in the strong a-DMRs (red dots). (F) Histograms illustrating the DNA content as assayed by propidium iodide (PI) staining in p51R cells that were cycling or blocked in G0/G1 for 4 days by incubation in a culture medium containing 0.5% fetal bovine serum. The G0/G1 block is almost complete. (G) Box plot illustrating CpG methylation in G0/G1 and in cycling (control) cells. Cells blocked in G0/G1 are statistically significantly more methylated than the cycling cells (P < .001, Wilcoxon rank sum test). (H) Bar plots illustrating the proportion of PMDs and HMDs in G0/G1 and in cycling (control) cells. Cells blocked in G0/G1 contain fewer PMDs than cycling cells.