eQTL Analysis

Genetic variation in a population is commonly studied through the analysis of single nucleotide polymorphisms (SNPs), which are genetic variants occurring at specific sites in the genome.   Expression quantitative trait loci (eQTL) analysis seeks to identify genetic variants that affect the expression of one or more genes: a gene-SNP pair for which the expression of the gene is associated with the allelic configuration of the SNP  is referred to as an eQTL.  Identification of eQTLs has proven to be a powerful tool in the study and understanding of diseases in human and other populations.

Using modern genotype and expression arrays, a typical eQTL analysis can involve millions  of SNPs and tens of thousands of genes, making computation and multiple testing key challenges.  Even local (cis) eQTL analyses that restrict attention to nearby genes and SNPs can involve tens of millions of gene-SNP pairs.  Our initial work on eQTL analysis addressed fast computation of association statistics in homozygous populations, and subsequent testing.  We are currently investigating the use of  iterative testing methods to enhance the power of full (trans) eQTL analyses.   Complementing eQTL testing, we have recently developed a simple log-of-linear model for assessing the effect size of an eQTL, an important problem that has not received much systematic attention in the literature.

To date, most eQTL studies have considered the effects of genetic variation on expression within a single tissue (typically blood).  An important next step is the simultaneous analysis of eQTLs  in multiple tissues.  Multi-tissue analysis has the potential to improve the findings of single tissue eQTL studies by borrowing strength across tissues, and to address fundamental biological questions about the nature and source of the differences between tissues.  An important feature of multiple tissue studies is that a SNP may be associated with the expression of a gene in some tissues, but not in others.    Working with the NIH Genotype-Tissue Expression (GTEx) Consortium, we developed  an empirical Bayes procedure, called MT-eQTL, for multi-tissue eQTL analysis.  The procedure, which is able to test for complex patterns of association across multiple tissues, was one of two methods used for testing eQTLs in the Consortium’s recent Science paper.  The MT-eQTL procedure is limited to nine or ten tissues, but we  are currently working on extensions that will scale to as many as twenty or thirty tissues.