How do you design and interpret epigenome-wide association studies (EWAS)?

EWAS design requires careful consideration of tissue relevance (target vs surrogate), sample size (effect sizes typically small), array vs sequencing platforms (Illumina EPIC for coverage, WGBS for resolution), and potential confounders. Analysis workflow includes quality control (detection p-values, beta value distribution), normalization (SWAN, functional normalization), batch effect correction (ComBat, SVA), statistical testing with multiple testing correction (FDR), and cell-type composition adjustment (reference-based deconvolution or RefFreeEWAS). Interpretation challenges include reverse causation (disease causing methylation changes), distinguishing cause from consequence, and functional validation. Replication in independent cohorts and integration with GWAS enhances findings.