This repository contains analysis scripts for our manuscript characterising differences in DNA methylation profiles between periperhal tissues and blood cell types, and how variation in blood cell types combines to influence variation in whole blood.
Most epigenome-wide association studies (EWAS) quantify DNA methylation (DNAm) in peripheral tissues such as whole blood to identify positions in the genome where variation is statistically associated with a trait or exposure. As whole blood comprises a mix of cell types, it is unclear whether trait-associated variation is specific to an individual cellular population. We collected three peripheral tissues (whole blood, buccal and nasal epithelial cells) from thirty individuals. Whole blood samples were subsequently processed using fluorescence-activated cell sorting (FACS) to purify five constituent cell-types (monocytes, granulocytes, CD4+ T cells, CD8+ T cells, and B cells). DNAm was profiled in all eight sample-types from each individual using the Illumina EPIC array. We identified significant differences in both the level and variability of DNAm between different tissues and cell types, and DNAm data-derived estimates of age and smoking were found to differ dramatically across sample types from the same individual. We found that for the majority of loci variation in DNAm in individual blood cell types was only weakly predictive of variance in DNAm measured in whole blood, however, the proportion of variance explained was greater than that explained by either buccal or nasal tissues. Instead we observe that DNAm variation in whole blood is additively influenced by a combination of the major blood cell types. For a subset of sites variable DNAm detected in whole blood can be attributed to variation in a single blood cell type providing potential mechanistic insight. We identified major differences in DNAm between blood cell types and peripheral tissues, with each sample type being characterized by a unique DNAm signature across multiple loci. Our results suggest that associations between whole blood DNAm and traits or exposures reflect differences in multiple cell types and provide important insights for the interpretation of EWAS performed in whole blood.