Fig. 3: DNA sequence variability affects CpG methylation and gene expression.

Illustration of the two models (models 1 and 2; bottom left) consistent with our results. In this hypothetical example, an ASM-QTL gives rise to CpG methylation differences between the maternal (left) and paternal (right) chromosomes of an individual. Under model 1, the ASM-QTL influences TF binding to DNA, which in turn influences methylation of nearby CpGs, but it is the TF (not methylation) that then results in influences on gene expression. Under model 2, the ASM-QTL influences TF binding to DNA, which again leads to influences on methylation of nearby CpGs, but here the change in methylation results in influences on gene expression, for example, by enabling binding of a CpG methylation-sensitive TF. Hence, methylation is irrelevant to gene expression in model 1 whereas it is relevant to gene expression in model 2. In both models, it is DNA sequence variability that drives the correlation between CpG methylation and gene expression. ALT, alternative allele; REF, reference allele; RNAPII, RNA polymerase II.