Correction for hidden confounders in the genetic analysis of gene expression

Listgarten, Jennifer; Kadie, Carl; Schadt, Eric E.; Heckerman, David

doi:10.1073/pnas.1002425107

Cited by 147 publications

(157 citation statements)

References 35 publications

Supporting

Mentioning

156

Contrasting

Order By: Relevance

“…We have shown here, for example, that reduction in gene expression measurement noise by PC analysis can markedly improve the ability to identify novel eQTLs. Similar approaches have previously been shown to improve power for eQTL mapping Choy et al 2008;Kang et al 2008;Listgarten et al 2010;Pickrell et al 2010;Stegle et al 2010). Compared to the linear mixed model analyses (Kang et al 2008;Listgarten et al 2010), our approach is more similar to surrogate variable analysis (SVA) or to the Bayesian factor analysis model (VBQTL) used by Stegle et al (2010) and Pickrell et al (2010), in that the unobserved confounders are modeled explicitly.…”

Section: Discussionmentioning

confidence: 73%

A cross-platform analysis of 14,177 expression quantitative trait loci derived from lymphoblastoid cell lines

et al. 2013

View full text Add to dashboard Cite

Gene expression levels can be an important link DNA between variation and phenotypic manifestations. Our previous map of global gene expression, based on ∼400K single nucleotide polymorphisms (SNPs) and 50K transcripts in 400 sib pairs from the MRCA family panel, has been widely used to interpret the results of genome-wide association studies (GWASs). Here, we more than double the size of our initial data set with expression data on 550 additional individuals from the MRCE family panel using the Illumina whole-genome expression array. We have used new statistical methods for dimension reduction to account for nongenetic effects in estimates of expression levels, and we have also included SNPs imputed from the 1000 Genomes Project. Our methods reduced false-discovery rates and increased the number of expression quantitative trait loci (eQTLs) mapped either locally or at a distance (i.e., in cis or trans) from 1534 in the MRCA data set to 4452 (with <5% FDR). Imputation of 1000 Genomes SNPs further increased the number of eQTLs to 7302. Using the same methods and imputed SNPs in the newly acquired MRCE data set, we identified eQTLs for 9000 genes. The combined results identify strong local and distant effects for transcripts from 14,177 genes. Our eQTL database based on these results is freely available to help define the function of disease-associated variants.

show abstract

Section: Discussionmentioning

confidence: 73%

A cross-platform analysis of 14,177 expression quantitative trait loci derived from lymphoblastoid cell lines

et al. 2013

View full text Add to dashboard Cite

show abstract

“…This has led to the development of a number of methods to control for underlying confounding factors (Leek and Storey 2007;Kang et al 2008;Listgarten et al 2010;Stegle et al 2010;Fusi et al 2012;Gagnon-Bartsch and Speed 2012). However, these methods generally cannot distinguish trans-eQTL hotspots from batch effects.…”

mentioning

confidence: 99%

The Dissection of Expression Quantitative Trait Locus Hotspots

et al. 2016

View full text Add to dashboard Cite

Studies of the genetic loci that contribute to variation in gene expression frequently identify loci with broad effects on gene expression: expression quantitative trait locus hotspots. We describe a set of exploratory graphical methods as well as a formal likelihood-based test for assessing whether a given hotspot is due to one or multiple polymorphisms. We first look at the pattern of effects of the locus on the expression traits that map to the locus: the direction of the effects and the degree of dominance. A second technique is to focus on the individuals that exhibit no recombination event in the region, apply dimensionality reduction (e.g., with linear discriminant analysis), and compare the phenotype distribution in the nonrecombinant individuals to that in the recombinant individuals: if the recombinant individuals display a different expression pattern than the nonrecombinant individuals, this indicates the presence of multiple causal polymorphisms. In the formal likelihood-based test, we compare a two-locus model, with each expression trait affected by one or the other locus, to a single-locus model. We apply our methods to a large mouse intercross with gene expression microarray data on six tissues. KEYWORDS eQTL; pleiotropy; multivariate analysis; data visualization; gene expression T HERE is a long history of efforts to map the genetic loci [called quantitative trait loci (QTL)] that contribute to variation in quantitative traits in experimental organisms, particularly to learn about the etiology of disease (Broman 2001;Jansen 2007). But it remains difficult to identify the genes underlying QTL (Nadeau and Frankel 2000). There has been much interest recently in measuring gene expression in disease-relevant tissues in QTL experiments as a way to speed the process from QTL to gene (Jansen and Nap 2001;Albert and Kruglyak 2015). The genetic control of gene expression is itself of great interest.Expression quantitative trait loci (eQTL) analysis attempts to find the genomic locations that influence variation in gene expression levels [messenger RNA (mRNA) abundances]. eQTL near the genomic location of the influenced gene are called local eQTL, and eQTL far away from the influenced gene are called trans-eQTL. When a genomic region influences the expression of many genes, the region is called a trans-eQTL hotspot.eQTL hotspots have been observed in many genetic studies (e.g., Brem et al. 2002;Schadt et al. 2003;Yvert et al. 2003;Chesler et al. 2005), and they are of particular interest because gene expressions mapping to the same location may indicate the existence of a genetic regulator.Batch effects (artifacts arising from technical or environmental factors) are common in microarray experiments. This has led to the development of a number of methods to control for underlying confounding factors (Leek and Storey 2007;Kang et al. 2008;Listgarten et al. 2010;Stegle et al. 2010;Fusi et al. 2012;Gagnon-Bartsch and Speed 2012). However, these methods generally cannot distinguish trans-eQTL hotspots ...

show abstract

“…On the other hand, spurious eQTL associations tend to inflate the discovery of trans-acting eQTLs (Breitling et al 2008). From this perspective, it makes sense to expect an enrichment of cis-eQTLs when indirect associations are effectively discarded (Kang et al 2008;Listgarten et al 2010).…”

Section: Qp-graph Estimates Of Eqtl Network Are Enriched For Cis-actmentioning

confidence: 99%

“…These problems can be addressed by estimating and including the confounding factors in the model as main (Leek and Storey 2007;Stegle et al 2010) or mixed effects (Kang et al 2008;Listgarten et al 2010) and restricting the eQTL search to cis-acting variants located in the regulatory regions of the gene to which they are associated (Montgomery et al 2010). …”

mentioning

confidence: 99%

“…They can be broadly categorized into those extending univariate single-marker regression models to adjust for confounding effects (Leek and Storey 2007;Stegle et al 2010;Listgarten et al 2010) and those using multivariate approaches. The latter can be further categorized into Bayesian networks using conditional mutual information with constraint-based algorithms (Zhu et al 2004), empirical Bayes hierarchical mixtures (Kendziorski et al 2006), directed versions of the PC algorithm (Chaibub Neto et al 2008), structural equation models (Liu et al 2008), sparse partial least squares (Chun and Keleş 2009), fused Lasso regression methods (Kim and Xing 2009), random forests (Michaelson et al 2010), mixed Bayesian networks using the Bayesian information criterion (BIC) with homogeneous conditional Gaussian regression models (Chaibub Neto et al 2010), mixed graphical Markov models restricted to tree network topologies (Edwards et al 2010), sparse factor analysis (Parts et al 2011), and conditional independence tests of order one (Bing and Hoeschele 2005;Chen et al 2007;Kang et al 2010;Chaibub Neto et al 2013).…”

mentioning

confidence: 99%

See 1 more Smart Citation

Mapping eQTL Networks with Mixed Graphical Markov Models

2014

View full text Add to dashboard Cite

Expression quantitative trait loci (eQTL) mapping constitutes a challenging problem due to, among other reasons, the highdimensional multivariate nature of gene-expression traits. Next to the expression heterogeneity produced by confounding factors and other sources of unwanted variation, indirect effects spread throughout genes as a result of genetic, molecular, and environmental perturbations. From a multivariate perspective one would like to adjust for the effect of all of these factors to end up with a network of direct associations connecting the path from genotype to phenotype. In this article we approach this challenge with mixed graphical Markov models, higherorder conditional independences, and q-order correlation graphs. These models show that additive genetic effects propagate through the network as function of gene-gene correlations. Our estimation of the eQTL network underlying a well-studied yeast data set leads to a sparse structure with more direct genetic and regulatory associations that enable a straightforward comparison of the genetic control of gene expression across chromosomes. Interestingly, it also reveals that eQTLs explain most of the expression variability of network hub genes.

show abstract

Correction for hidden confounders in the genetic analysis of gene expression

Cited by 147 publications

References 35 publications

A cross-platform analysis of 14,177 expression quantitative trait loci derived from lymphoblastoid cell lines

A cross-platform analysis of 14,177 expression quantitative trait loci derived from lymphoblastoid cell lines

The Dissection of Expression Quantitative Trait Locus Hotspots

Mapping eQTL Networks with Mixed Graphical Markov Models

Contact Info

Product

Resources

About