EpipwR: efficient power analysis for EWAS with continuous outcomes

Epigenome-wide association studies (EWAS) have emerged as a popular way to investigate the pathophysiology of complex diseases and to assist in bridging the gap between genotypes and phenotypes. Despite the increasing popularity of EWAS, very few tools exist to aid researchers in power estimation and those are limited to case-control studies. The existence of user-friendly tools, expanding power calculation functionality to additional study designs, would be a significant aid to researchers planning EWAS.

We introduce EpipwR, an open-source R-package that can efficiently estimate power for EWAS with continuous or binary outcomes. EpipwR uses a quasi-simulated approach, meaning that data is generated only for CpG sites with methylation associated with the outcome, while P-values are generated directly for those with no association (when necessary). Like existing EWAS power calculators, reference datasets of empirical EWAS are used to guide the data generation process. Two numerical studies show the effect of the selected empirical dataset on the generated correlations and power, while another explores the accuracy of EpipwR against case-control alternatives. EpipwR is shown to outperform existing alternatives on both simulated and real EWAS datasets.

The EpipwR R-package is currently available on Bioconductor or at github.com/jbarth216/EpipwR.