Guidance for the design and analysis of cell-type specific epigenetic epidemiology studies

Recent studies on the role of epigenetics in complex disease have focused primarily on DNA methylation. As genome-wide patterns of DNA methylation are largely determined by cell type, analyses of bulk tissues, composed of a heterogeneous mix of cell types, limits the ability to identify both the specific loci affected by disease associated differences and the cell type within which these changes occur. Cell-specific datasets now make it possible to identify DNA methylation differences associated with disease in specific cell types. Critically, the analytical framework for these differ from the traditional bulk tissue approach, as the statistical method needs to characterise whether the difference affects multiple cell types or is specific to a particular cell-type. In this manuscript, we evaluate the effects on study design, data preprocessing and statistical analysis for cell-specific studies of DNAm, particularly for scenarios where multiple cell types are considered. For this study we take advantage of a large set of DNA methylation profiles (n = 751) obtained from five different purified cell populations isolated from human prefrontal cortex nuclei samples. Post-mortem tissue from 287 adult donors (aged 18-108 years old) were processed using a FANS protocol developed by our group (Policicchio et al. 2020) into up to four purified populations of cells per individual. After stringent quality control of the DNA methylation data 751 nuclei samples were retained: 218 Total, 164 DoubleNeg, 182 NeuNPos, 168 Sox10Pos, 12 IRF8Pos and 7 TripleNeg nuclei samples.