We showed in our study  that SNP rs10876043 in the disco-interacting protein 2 homolog B gene (DIP2B) was associated with the first principal component of methylation. Although the analyses and result remain unchanged, it appears that this observation is likely due to a genotyping artifact. That is, the reported rs10876043 genotypes differ according to HapMap Phase (cell lines genotyped in Phase 1/2 have reported genotypes AG and GG, while Phase 3 cell lines have genotype AA). The 1000 Genomes data suggest the correct genotype is probably AA for all of these YRI individuals. These genotype differences between different phases of the HapMap Project, coupled with a small difference in mean methylation between Phase 1/2 vs 3 cell lines appear to have produced an artifactual association. Other analyses in the paper controlled for the top principal components and should therefore be robust to this type of effect.
We thank Lude Franke and Harm-Jan Westra (Department of Genetics, University Medical Centre Groningen) for bringing this to our attention.