Fig. 3

meQTL analysis. The red line represents cases where no SNP-CpG association was detected at 95% power within the given window. The black line represents cases where a SNP-CpG association was found at a 95% certainty of not falsely rejecting the null hypothesis. The actual number of SNPs tested in the respective window was used to correct for multiple testing (Bonferroni correction), effectively limiting this analysis to a maximal distance of 5000 bp from the CpG site, above which the presence of any SNP-CpG association can be neither confirmed nor rejected with 95% certainty. The region between 0–110 bp corresponds to the probe-binding region. CpGs that contain SNPs in this region were excluded from the analysis. For instance, at 95% power, only 16% of the sites have a SNP within the first 100 nucleotides from the end of the probe-binding region (210 bp from the CpG), while 54% of the sites have a SNP within the first 1000 nucleotides. In KORA, only genotyped SNPs are available; hence, a windowed approach was not possible. However, the KORA study is sufficiently powered to detect any SNP association inside a window of 5 Mb, revealing that 97% of the sites have a SNP-CpG association in that region. The dashed line represents a likely extrapolation of the pattern that would be expected in a similar windowed analysis in KORA