Meta-analysis of genomic regions in epigenome-wide association studies
Year: 2020
Type: review
Abstract: Meta-analysis is a commonly used strategy for synthesizing information to identify consistent effects in multiple studies. Epigenome-wide association studies (EWAS) examine association between DNA methylation markers in the epigenome with phenotypes or diseases. Biologically, it has been observed that methylation levels are strongly correlated across the genome and methylation often occurs as a regional phenomenon. Statistically, while effect size in a single CpG might be small, by borrowing information from all the CpGs within a region, statistical test for regions can more effectively leverage information within the region to increase sensitivity and specificity. However, to date, while a number of tools have been developed to identify differentially methylated regions within a single study, most meta-analyses have been performed on individual CpGs. This is because significant regions identified in different datasets are data dependent and typically vary across datasets. DMR (differentially methylated regions) identification methods typically start with computing p-values for differential methylation at individual CpG sites, and then scan the genome to identify regions with adjacent low p-values. Therefore, DMRs identified by these methods can be overlapping but often vary in different studies. In addition, the analysis results typically only include information on DMRs, the lack of information on the rest of the genome, which might include regions with moderate changes, albeit not reaching statistical significance, makes it difficult for meta-analysis. To address this challenge, we introduce two analyses pipelines for conducting meta-analysis for genomic regions in EWAS, using comb-p software and coMethDMR R package. We illustrate, compare and contrast the results of these two approaches by performing meta-analysis for several Alzheimer’s Disease brain samples datasets.
Institutions: University of Miami, Sylvester Comprehensive Cancer Center
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: gold
OpenAlex ID: W3161969796