A Robust Statistical Framework for eQTL Analysis

• Activity Number: 677
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Mohamedou Sow* and Gilles Durrieu and Laurent Briollais+
• Companies: Samuel Lunenfeld Research Institute and University Bordeaux 1 and Samuel Lunenfeld Research Institute
• Keywords: genomics ; non-parametric ; regression ; quantile ; sib-pair

Abstract

Combining global gene-expression profiling and genetic analysis of natural allelic variation (genetical genomics or eQTL analysis) has great potential in dissecting the genetic pathways underlying complex phenotypes. In general, the identification of biologically interesting genes in this context is challenging and complicated by high levels of experimental noise, the presence of outliers and non-normal distributions of the gene expressions. We propose a general framework based on quantile regression as a robust alternative to standard regression models to circumvent these problems. We further extent this framework to the analysis of paired data in particular for sib-pair designs and for investigation of gene correlations. We finally illustrate the interest of the method through an application to the human CEPH data.