Simulations allow estimation of the overall fraction of single nucleotide polymorphisms (SNPs) across the genome with differential local methylation. (a) Under different simulations, we calculated the sum of squares of fractional deviations of the simulated data compared with empirical data examining fractional differences in four parameters: methylation overlap, opposite overlap and methylation overlap in three individuals, and methylation overlap in four individuals. (b) The fractions of amplicons showing methylation overlap in groups of two, three and four individuals for observed (Obs.), expected (Exp.) and simulated (Sim.) data.

The distribution of the distances between the single nucleotide polymorphisms (SNPs) and the methylation-sensitive restriction enzyme (MSRE) sites present on amplicons displaying methylation overlap. (a) Generic amplicon with a single MSRE site. (b) Percentages of amplicons with given distances between the MSRE and the SNP, for amplicons with one MSRE site; 50 bp bins. See legend. (c) Allelic methylation patterns were analyzed in 10 individuals from three generations (grandparents, parents, children) using array-based methylation assessment (both genotyping and raw intensity assays), and bisulfite sequencing. For the three SNPs shown, the parents were homozygous for opposite alleles and all the children were heterozygous. Array-based genotyping data before and after MSRE treatment are separated by arrows. When the genotype call was homozygous after MSRE treatment, we denoted it as A or B. For the parents and grandparents, when the genotypes were all homozygous (for these SNPs) the genotype after MSRE treatment is displayed as AA or BB, reflecting the likelihood that both alleles were methylated, but the genotype calling would not itself reveal if one allele were unmethylated. Representative bisulfite sequencing results in one individual from each of three generations for the SNP rs1038492. The linked MSRE is underlined and the SNP is indicated with an asterisk. The 'A' allele is consistently methylated both at the SNP CpG and at the MSRE CpG. The 'B' allele is unmethylated at the MSRE site, so its genotype tends to disappear after MSRE treatment; upon bisulfite treatment, the 'B' allele has the C that is part of the CpG within the MSRE completely converted to T. (d) For SNPs in which one allele is a CpG, the SNP CpG was itself methylated along with the linked CpG within the MSRE.

Genome-scale assessment of allelic DNA methylation patterns. (a) Representative amplicon (defined by NspI sites). The 'B' allele has a methylated CpG (red filled oval) within a methylation-sensitive restriction enzyme (MSRE). The 'A' allele is unmethylated at the CpG within the MSRE site, and is therefore digested upon the treatment with the MSRE cocktail that is performed before NSPI digestion ligation of adaptors, PCR amplification and hybridization to the Affymetrix 250 K single nucleotide polymorphism (SNP) mapping array. Only the 'B' allele yields a PCR product that can hybridize on the SNP-detecting array. (b) Of the 8,346 SNPs heterozygous in two genetically unrelated individuals (GM10848 and GM10849), epigenotypes are shown, with the numbers of amplicons with monoallelic methylation in bold. The lower part of the panel shows the number of amplicons displaying methylation overlap and the number displaying opposite overlap. The P value was calculated using the χ² test; the null hypothesis was equality in the numbers of overlapping SNPs with the opposite or same epigenotype. (c) The observed methylation overlap frequencies in excess of the levels expected by chance are presented for various types of comparisons of pairs of individuals.