Cell-Type Heterogeneity in DNA Methylation Studies: Statistical Methods and Guidelines

Studies in epigenetic epidemiology have reported increasing numbers of epigenetic biomarkers associated with a wide range of exposures and outcomes. Due to cost and technical difficulties, these markers are usually derived from complex tissues that are composed of many different cell-types. This cell-type heterogeneity prevents the identification of cell-type specific epigenetic alterations, posing significant challenges to the interpretation and understanding of these markers. Consequently, there is a strong need to develop cost-effective computational solutions to tackle the cell-type heterogeneity problem. Here, I discuss some recently proposed cell-type deconvolution algorithms aimed at estimating cell-type fractions and identifying cell-type specific differential DNA methylation changes. I describe their successful application to epigenome studies. We also discuss their main limitations, providing general guidelines for their successful implementation and for correctly interpretating results derived from them.