ABSTRACT: (1) In clinical practice and biomedical research populations are often divided categorically into distinct racial and ethnic groups. In reality, these categories comprise diverse groups with highly heterogeneous histories, cultures, traditions, religions, as well as social and environmental exposures. While the factors captured by these categories contribute to clinical practice and biomedical research, the use of race/ethnicity is widely debated. As a response to this debate, genetic ancestry has been suggested as a complement or alternative to this categorization. However, few studies have examined the effect of genetic ancestry, racial/ethnic identity, and environmental exposures on biological processes. Herein, we examine the contribution of self-identification within ethnicity, genetic ancestry, and environmental exposures on epigenetic modification of DNA methylation, a phenomenon affected by both genetic and environmental factors. We typed over 450,000 variably methylated CpG sites in primary whole blood of 573 individuals of Mexican and Puerto Rican descent who also had high-density genotype data. We found that methylation levels at a large number of CpG sites were significantly associated with ethnicity even when adjusting for genetic ancestry. In addition, we found an enrichment of ethnicity-associated sites amongst loci previously associated with environmental and social exposures. Interestingly, one of the strongest associated sites is driven by the Duffy Null blood type variant, demonstrating a new function of the locus in lymphocytes. Overall, the methylation changes associated with race/ethnicity, driven by both genes and environment, highlight the importance of measuring and accounting for both self-identified race/ethnicity and genetic ancestry in clinical and biomedical studies and the benefits of studying diverse populations. (2) In epigenome-wide association studies (EWAS), different methylation profiles of distinct cell-types may lead to false discoveries. We introduce ReFACTor, a method based on principal component analysis (PCA) for the correction of cell-type heterogeneity in EWAS. ReFACTor does not require knowledge of the cell counts, and it obtains improved estimates of the cell-type composition, resulting in improved power and control for false positives in EWAS.
