Detection of Cell-Type-Specific Differentially Methylated Regions with Aggregate-Level DNA Methylation Data

• Activity Number: 537 - Innovative Statistical Methods for Complex -Omics Data
• Type: Topic Contributed
• Date/Time: Thursday, August 6, 2020 : 1:00 PM to 2:50 PM
• Sponsor: International Chinese Statistical Association
• Author(s): Yingying Wei*
• Companies: The Chinese University of Hong Kong
• Keywords: Epigenome-wide association studies; deconvolution; aggregate-level data; fine-scale inference

Abstract

Epigenome-wide association studies (EWAS) aim to identify the associations between CpG sites and phenotypes. However, the obtained methylation levels for each sample in an EWAS are actually signals aggregated from different cell types. Recently, there is emerging research on identifying cell-type-specific risk CpG sites with the aggregate-level DNA methylation data. However, although the power of association detection at the aggregate-level has significant improvements, the accurate detection of cell-type-specific risk CpG sites still asks for large sample sizes. Here, we propose a hierarchical model to detect genomic regions, instead of individual CpG sites, whose methylation levels are associated with phenotypes for each individual cell type. Our proposed models borrow strengths from nearby CpG sites and improve statistical power.