CMStatistics 2022: Start Registration
View Submission - CMStatistics
B1660
Title: Correlation-adjusted simultaneous testing among small-sized groups in high-dimensional DNA methylation data Authors:  Patrick Wincy Reyes - University of the Philippines (Philippines) [presenting]
Iris Ivy Gauran - King Abdullah University of Science and Technology (Saudi Arabia)
Erniel Barrios - University of the Philippines (Philippines)
Hernando Ombao - King Abdullah University of Science and Technology (KAUST) (Saudi Arabia)
Abstract: Epigenetics plays a crucial role in understanding the underlying molecular processes of Type 2 Diabetes (T2D) and determining therapeutic targets. In a natural experiment of life conducted on T2D to investigate the complex interplay of genetic variation compared to environmental exposures, we aim to identify differentially methylated probes (DMPs) between controls and cases. Statistically, this is a high-dimensional testing problem with sparse signals and correlated variables across an inherent grouping structure. We propose a class of multiple testing procedures that utilizes the correlation within the genes to control the False Discovery Rate (FDR). Simulation studies show that the proposed methods have superior empirical power while controlling the FDR compared to the benchmark procedures such as Group Benjamini-Hochberg and Group Benjamini-Yekutieli methods. These existing methods fail to control the FDR when the data is grouped with correlated probes. We applied the methods to the data containing a sample of 346 Filipinos enrolled in either Manila or California. Using p-values from the analysis with covariates, we identified a much lower number of significant DMPs, which may facilitate more cost-efficient experimental studies for scientists in identifying novel therapeutic tools for the treatment of Diabetes Mellitus.