Epigenome-wide association studies for common human diseases

Abstract

Epigenetic variation affects genome function and hence can contribute to common disease. To establish a possible link requires systematic studies, such as the proposed epigenome-wide association studies (EWASs). Of the many epigenetic marks, DNA methylation (DNAm) is the most stable and accessible and therefore ideally suited for EWASs. In principle, EWASs should be equally successful as genome-wide association studies (GWASs) for the identification of disease-associated variations. However, there are fundamental differences between GWASs and EWASs that need to be considered for appropriate study design. The key differences for EWASs are tissue specificity and the possibility that some epigenetic changes may occur downstream of the disease process. Both considerations affect the type of cohorts and samples that should be analyzed. Technologies for EWASs are readily available for both array- and sequencing-based platforms but many of the computational and statistical analysis methods remain to be developed. At this early stage, it is challenging to predict the possible effect of DNAm variation. However, if it does exist and if the right study design is used, then much more than the 'low-hanging fruit' should be detectable in fewer samples than are required for a typical GWAS, based on simulations assuming a conservative methylation odds ratio.