Abstract

Efforts to identify the genetic loci responsible for variation in quantitative traits have traditionally focused on one or at most a few phenotypes. With high-throughput technologies now widely available, investigators can measure thousands of phenotypes for quantitative trait loci (QTL) mapping. Gene expression measurements are particularly amenable to QTL mapping and the results from these expression QTL (eQTL) studies have proven utility in addressing a number of important biological questions. Although useful in many ways, the results are limited by lack of statistical methods designed specifically for this problem. Most studies to date have applied single QTL trait analysis methods to each expression trait in isolation. Doing so can reduce the power for eQTL localization since information common across transcripts is not utilized; furthermore, false discovery rates can be inflated if relevant multiplicities are not considered. To maximize the information obtained from eQTL mapping studies, new statistical methods are required. We here review the eQTL mapping problem and commonly used approaches, and we propose a new method to facilitate eQTL interval mapping. Results are demonstrated using simulated data and data from a study of diabetes in mouse.

Introduction

Although efforts to identify the genetic loci responsible for variation in quantitative traits have been going on for over 80 years, the vast majority of studies have taken place in the last two decades. This is due largely to two major advances in the 1980s: the advent of restriction fragment length polymorphisms (RFLPs) making it possible to genotype markers on a large scale and the advent of statistical methods for the related data analysis.