A0238
Title: High-dimension to high-dimension screening for detecting genome-wide epigenetic regulators of gene expression
Authors: Tianzhou Ma - University of Maryland (United States) [presenting]
Abstract: The advancement of high-throughput technology characterizes a wide range of epigenetic modifications across the genome that regulate gene expression. The high dimensionality of both epigenetic and gene expression data make it challenging to identify the epigenetic regulators of genes over the whole genome. Conducting a univariate test for each epigenetic-gene pair is subject to serious multiple comparison burden, and direct application of regularization methods to select important epigenetic-gene pairs is computationally infeasible for both high-dimensional predictors and responses. Applying fast screening to reduce the dimension first before regularization is more efficient and stable than applying regularization methods alone. We propose a high-dimension to high-dimension screening method based on robust partial correlation, namely rPCor, in a multivariate regression model for detecting epigenetic regulators of gene expression over the whole genome. Compared to existing screening methods, our method can reduce the dimension of both predictor and response, and screen at both node (epigenetic features or genes) and edge (epigenetic-gene pairs) levels. We develop data-driven procedures to determine the conditional sets and the optimal screening threshold and implement a fast iterative algorithm. Simulations and two real data applications in cancer studies illustrate the validity and advantage of our method.
