Epigenome-wide association study (EWAS)

An epigenome-wide association study (EWAS) is a hypothesis-free scan that tests associations between DNA methylation levels at hundreds of thousands of CpG sites (cytosine–guanine dinucleotides where the cytosine can acquire a methyl group) across the genome and a phenotype of interest — such as chronological age, a disease, or an environmental exposure. Methylation is typically quantified using microarrays (most commonly the Illumina HumanMethylation450 or EPIC/850K BeadChip) or whole-genome bisulfite sequencing, yielding beta-values between 0 and 1 for each site; a linear or logistic regression is then run at every CpG, with correction for multiple testing and for confounders including estimated blood-cell-type proportions. In aging research, EWAS generated many of the CpG training sets that underlie first-generation epigenetic clocks: Hannum et al. (2013) used methylation data from 656 blood samples spanning ages 19–101 to identify 71 age-associated CpGs that predicted biological age with high accuracy in independent cohorts, and similar approaches led to Horvath's 353-CpG pan-tissue clock. The EWAS Catalog (Battram et al., 2022) aggregates over 1.7 million associations from more than 2,600 EWAS — including both peer-reviewed publications and unpublished scans — enabling lookup of CpG-phenotype links across cohorts. A persistent methodological challenge is reverse causation: a disease or age-related process can itself reshape the methylome, so an observed association does not establish that the CpG change precedes or drives the outcome; Mendelian randomisation and longitudinal designs are increasingly used alongside EWAS to triangulate causal direction.