An integrative genomics approach to infer causal associations between gene expression and disease

Abstract

A key goal of biomedical research is to elucidate the complex network of gene interactions underlying complex traits such as common human diseases. Here we detail a multistep procedure for identifying potential key drivers of complex traits that integrates DNA-variation and gene-expression data with other complex trait data in segregating mouse populations. Ordering gene expression traits relative to one another and relative to other complex traits is achieved by systematically testing whether variations in DNA that lead to variations in relative transcript abundances statistically support an independent, causative or reactive function relative to the complex traits under consideration. We show that this approach can predict transcriptional responses to single gene-perturbation experiments using gene-expression data in the context of a segregating mouse population. We also demonstrate the utility of this approach by identifying and experimentally validating the involvement of three new genes in susceptibility to obesity.In the past few years, gene-expression microarrays and other general molecular profiling technologies have been applied to a wide range of biological problems and have contributed to discoveries about the complex network of biochemical processes underlying living Correspondence should be addressed to E.E.S. (eric_schadt@merck.com). Note: Supplementary information is available on the Nature Genetics website. COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests. NIH Public Access Author ManuscriptNat Genet. Author manuscript; available in PMC 2010 March 18. Published in final edited form as:Nat Genet. 2005 July ; 37(7): 710-717. doi:10.1038/ng1589. NIH-PA Author ManuscriptNIH-PA Author Manuscript NIH-PA Author Manuscript systems 1 , common human diseases 2,3 and gene discovery and structure determination [4][5][6] . Microarrays have also helped to identify biomarkers 7 , disease subtypes 3,8,9 and mechanisms of toxicity 10 and, more recently, to elucidate the genetics of gene expression in human populations 11,12 and to reconstruct gene networks by integrating gene-expression and genetic data 13 . The use of molecular profiling technologies as tools to identify genes underlying common, polygenic diseases has been less successful. Hundreds or even thousands of genes whose expression changes are associated with disease traits have been identified, but determining which of the genes cause disease rather than respond to the disease state has proven difficult.Microarray data have recently been combined with other experimental approaches to facilitate identification of key mechanistic drivers of complex traits 3,[13][14][15][16][17] . One such technique involves treating relative transcript abundances as quantitative traits in segregating populations. In this method, chromosomal regions that control the level of expression of a particular gene are mapped as expression quantitative trait loci (eQTLs). Gene-expression QTLs that contain the gene encoding t...

The Diasporin Pathway: a tumor progression-related transcriptional network that predicts breast cancer survival

“…We identified three loci on chromosomes 7, 17 and 18 that appear to be linked to metastasis-predictive ECM gene expression. Subsequently, correlation analysis was performed to facilitate candidate gene identification since the current prevailing hypothesis is that most modifiers are likely to result from modest variations in gene expression levels or mRNA stability [23,24]. This revealed a correlation between expression of a variety of genes within the peak region of linkage of each eQTL and the expression of various metastasis-predictive ECM genes (Supplementary Table 4 & Supplementary Table 5 and Crawford et al [10]).…”

“…Correlation analysis was performed to identify potential candidates for the chromosomes 7, 17 and 18 ECM modifiers since the current prevailing hypothesis that most modifiers are likely to result from modest variations in gene expression levels or mRNA stability [23,24]. Whole genome correlation analysis of the microarray data was performed using the Trait Correlation function of the WebQTL database within the GeneNetwork web service [18] to identify probe sets with high correlation coefficients with each of the ECM genes of interest.…”

Obesity and polymorphisms in genes regulating human adipose tissue

“…For example, by combining mapping of adiposity phenotypes with liver expression profiles in an F2 intercross, Zfp90, C3ar1 and Tgfbr2 were identified as causal for obesity. 12 From such studies, it is clear that some expression quantitative trait loci (QTLs), that is genetic loci regulating specific gene expression, colocalize with disease QTLs supporting that differentially expressed genes are causal for disease. 13 Schadt et al 13 have concluded that genes whose expression levels significantly correlate with a complex phenotype, transcript abundance are controlled by QTLs co-localizing with phenotypic QTLs and physical location are supported by phenotypic and expression QTLs are natural causal candidates for disease.…”

Genome-wide system analysis reveals stable yet flexible network dynamics in yeast

“…To determine the degree of modularity and to identify the modules, we apply a random walk Markov CLustering algorithm (MCL) 5 [34,35] to the symmetric version, ) (s w , of the weight matrix, w. 6 The present network turns out to be highly modular (Modularity = 0.74 [37] where n is the number of modules. This reveals the global P-value of the graph theoretic modules being associated to coherent biological processes to be less than 5 10  , thus biologically validating the inferred modular architecture. More specifically, several (17) modules contain significant groups of genes involved in the same processes (P k < 0.01), e.g., biosynthesis, ribosome biogenesis and DNA replication.…”

“…This data is often analyzed by clustering over different experiments of wholegenome expression profiles, and that technique has provided important insights into gene function [2]. However, clustering alone cannot resolve gene interactions, and progress in network identification algorithms has revealed aspects of the static wiring of gene networks [3][4][5][6][7][8][9][10][11]. A recent study by Luscombe and colleagues [8] provided a first step towards an understanding of network dynamics by describing when different sub-networks are active during different cellular conditions in Yeast.…”