This version corresponds to the one proposed in Jumentier et al. 2023. "High-Dimensional Mediation Analysis: A New Method Applied to Maternal Smoking, Placental DNA Methylation, and Birth Outcomes"
-
Step 1. Evaluating associations between exposure, mediators and outcome, using a latent factor mixed models, using "mEWAS" function
-
Step 2. Identifying potential CpG mediators and aggregated mediator regions, combining the significance values Px and Py of each previous association to compute a P-value for each CpG, with "max2" function.
-
Step 3. Quantifying indirect effects with single and multiple mediators with "wrap_mediation" and "est_oie" functions
-
Extra functions allow to use AMR (aggregated mediator regions) instead of single CpG with "combp2", "AMR_search" and "AMR_build" functions