Zoom: https://iastate.zoom.us/j/98987943931?pwd=RisrdVRwZHErbVdVeERQMithbk1sdz09

Presenter: Amin Shirazi, Ph.D. Candidate in Statistics

Title: Power Enhanced Gene Differential Expression Analysis by Incorporating Gene Network Information

Abstract: Detecting differentially expressed (DE) is a fundamental step in gene expression data analysis. Most methods for identifying DE genes ignore the dependence structure among genes, and such procedures may lose power when there is a strong dependence among genes. We propose a new testing procedure called the Dependence Boosted Differential Expression (DBDE) procedure, which incorporates the information of the gene covariance structure through linear models, and hence utilizes the dependence structure that can be obtained from prior biological knowledge or gene networks. Our test statistic is based on the residuals of the linear models and has a higher signal-to-noise ratio, leading to more powerful tests. Through simulation studies, we show the proposed DBDE procedure improves power for identifying DE genes when there is moderate to strong dependence on genes and under various dependence structures. Our simulation studies also indicate that the DBDE can control the false discovery rate at any nominal level. We also construct a computation pipeline for implementing the proposed methods in real data analysis.