Fig. 1: Overview of EWASplus approach.

The EWASplus procedure is composed of four major steps: (1) Training data collection from existing EWASs; (2) External feature (from sources such as ENCODE and Roadmap Epigenome consortia) selection; (3) Ensemble learning; and (4) Genome-wide CpGs risk prediction, in which trained ensemble learning model is applied genome-wide to score all CpGs.