Unlock your biological data
Recent work applying EWAS suggests an important role for DNA methylation as a mechanism involved with disease. In a standard EWAS of primary tissue such as whole-blood, methylation data represent the epigenetic states of a heterogeneous mixture of cell-types. Since the epigenome is highly variable across different cell-types, correlations between the phenotype of interest and the cell-type composition lead to a large number of false discoveries(Jaffe et al.,2014; Zou et al., 2014).
The standard statistical analysis applied in EWAS uses a univariate test for correlation between the phenotype and each of the probed CpG sites. Thus, false discoveries due to cell-type heterogeneity can be addressed by adding the cell proportions as covariates. However, cell-type compositions are typically not measured and therefore a computational method has been proposed for the estimation of cell-type composition using a reference dataset which includes methylation measurements for sorted cells (Houseman et al., 2012).
(Jaffe et al.,2014) Accounting for cellular heterogeneity is critical in epigenome-wide association studies. Genome Biol.
(Zou et al., 2014) Epigenome-wide association studies without the need for cell-type composition. Nat Methods.
(Houseman et al., 2012) DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics.
(Rahmani et al., 2016) Sparse PCA corrects for cell type heterogeneity in epigenome-wide association studies. Nat Methods.