A novel three-class ROC method for eQTL analysis

Abstract

The problem of identifying genetic factors underlying complex and quantitative traits such as height, weight and disease susceptibility in natural populations has become a major theme of research in recent years. Aiming at revealing the inter-dependency and causal relationship between the underlying genotypes and observed phenotypes, researchers from different areas have developed a variety of methods for expression quantitative trait loci (eQTL) mapping. Most of these methods rely on resampling-based algorithms that are computationally very expensive. To overcome the disadvantages of the current techniques, we propose a novel nonparametric method based on the volume under surface (VUS) within the framework of three-class receiver operating characteristic (ROC) analysis. With the fast algorithms developed, we can reduce the computation time of the genomewide analysis from several months down to several days. © 2010 IEEE.