We annotated discovered eVariants using chromatin state predictions from 128 cell types or cell lines sampled by the Roadmap Epigenomics project2. Genome segmentation was performed for each cell type or cell line using a 15-state hidden Markov model (HMM) over 400 bp windows. Several of the learned states are labelled as enhancers, promoters, and repressed regions. For the standard 15-state Roadmap segmentations, regulatory elements are labelled independently for each cell type. For enrichment analyses, we constructed background variants sets that matched eVariants to randomly selected variants based on chromosome, distance to nearest TSS, and MAF.
