A Bayesian Approach to Analyzing Differential Gene Expression in Heterogeneous Tissue Samples

• Activity Number: 178 - Statistical Methods for Analysis of Heterogeneous Tissue Samples in Bulk and Single-Cell Sequencing Data
• Type: Contributed
• Date/Time: Monday, July 30, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Megan Stefanski* and David Spade
• Companies: University of Missouri - Kansas City and University of Missouri - Kansas City
• Keywords: Bayesian; Heterogeneous Tissue; Cancer; RNA-seq; Microarray; Gene Expression

Abstract

Identification of cancer biomarkers can aid in determining personalized treatment plans and improve patient outcomes. RNA transcript quantification of cancer tissue helps to identify biomarkers by determining which genes are over­-or under-­expressed, leading to aberrant cell behavior. Because tumor samples are inherently heterogeneous and manual separation is time consuming and costly, a new Bayesian method for mixed tissue expression analysis has been developed. This method utilizes microarray expression data and prior information on constituent tissue proportions to determine individual tissue­ specific expression. This method is currently being adapted to analyze RNA-seq data, a newer format of expression data, and utilize methylation data to have more informative priors on tissue proportion. This work presents the results of differential gene expression analysis in heterogeneous tissue and evaluates effects of new data with the model, specifically in regard to convergence and computational efficiency.