Incorporating prior information into association studies

Genome-wide association studies (GWAS) seek to identify genetic variants involved in specific traits. GWAS are advantageous for linking variants with traits, because they interrogate the genome in a uniform way. In other words, they examine the whole genome without a preconceived notion of where the associations may lie.

However, we now know a lot about the putative function of genetic variants due to tremendous progress in functional genomics. In many cases, we even know which variants are more likely to be involved in disease when compared to others. Advancements in our understanding of functional genomics motivate the strategic incorporation of prior information in GWAS.

Our group has been interested in this problem for many years. One challenge to addressing this problem is that the widely utilized approach for GWAS involves evaluating an association statistic at each single nucleotide polymorphism (SNP), and these methods take into account only one SNP at a time. The results are then adjusted for multiple testing, and an association is identified if a statistic exceeds a certain threshold. This approach can be described as a frequentist approach. On the other hand, one can incorporate prior information on which SNPs are likely to be the causal variants affecting the trait. This approach is inherently a Bayesian concept. Reconciling these two approaches is not straightforward.

Average power under varying relative risks. For more information, see our paper.

In a 2008 paper published in Genome Research, our group proposed a modification of the multiple testing framework to address this problem. Instead of using the same specific threshold for all of the association statistics, we use a different threshold for each association statistic, where the thresholds are adjusted based on the prior information. Our method takes advantage of the correlation structure by considering multiple markers within a region. In our paper, we demonstrate how to set the thresholds in order to optimally utilize prior information and maximize statistical power.

Using prior information in genetic association studies increases power over traditional association studies while maintaining the same overall false-positive rate. Compared to standard methods, our approach is equally simple to apply to association studies, produces interpretable results as p-values, and is optimal in its use of prior information in regards to statistical power.

In 2012, we extended this work to use only tag SNPs for the putative causal variant. This project was developed by Gregory Darnell (then UCLA undergraduate, now PhD student at Princeton University), Dat Duong (then UCLA undergraduate, now UCLA PhD student), and Buhm Han.

More recently, we have applied this framework to incorporate functional information in analysis of eQTL data. In this case, incorporating genomic annotation of variants significantly increases the statistical power of existing eQTL methods and detects more eGenes in comparison to standard approaches. Read the blog post on this paper, and download the full article.

For more information on our general approach, see our paper, which is available for download through Bioinformatics:
In addition, the open source implementation of our 2012 paper, MASA, which was developed by Greg Darnell and Dat Duong, is freely available for download at

The full citations to our papers on this topic are:

Darnell, Gregory; Duong, Dat ; Han, Buhm ; Eskin, Eleazar

Incorporating prior information into association studies. Journal Article

In: Bioinformatics, 28 (12), pp. i147-i153, 2012, ISSN: 1367-4811.

Abstract | Links | BibTeX

Eleazar Eskin. “Increasing Power in Association Studies by using Linkage Disequilibrium
Structure and Molecular Function as Prior Information.” Genome Research.
18(4):653-60 Special Issue Proceedings of the 12th Annual Conference on Research
in Computational Biology (RECOMB-2008), 2008.

Hypothalamic transcriptomes of 99 mouse strains reveal trans eQTL hotspots, splicing QTLs and novel non-coding genes

In a recent project, Farhad Hormozdiari and Eleazar Eskin contributed data analysis and interpretation to a project identifying new genes and genomic regions associated with metabolic function in mice. Our paper presents a comprehensive picture of the transcriptome of the mouse hypothalamus and its genetic variation and regulation. This project, which was published in eLife, was led by fellow UCLA researchers Yehudit Hasin-Brumshtein, Jake Lusis, and Desmond Smith.

Mice and humans share virtually the same set of genes; thus, mapping the mouse genome is an important step toward understanding genetic factors in common, complex human diseases such as obesity, heart disease, and diabetes. In metabolic tissues, the integration of genome-wide expression profiles with genetic and phenotypic variance can provide valuable insight into a disease’s underlying molecular mechanism. Measuring gene activity can reveal new molecules that clinical translation efforts may target to treat metabolic disorders.

Our project uses RNA-Seq to characterize transcriptome in 99 inbred strains of mice from the Hybrid Mouse Diversity Panel (HMDP), a reference resource population for cardiovascular and metabolic traits. Mice were fed a high, high sugar diet, and all strains were comprehensively genotyped and phenotyped for 150 metabolic traits. Our study examines tissues relevant to the hypothalmus, the brain region that controls metabolism and regulates body weight and appetite.

We sequenced 285 samples from all 99 strains of the HMDP. Using methods described in our paper, we identified thousands of new isoforms and >400 new genes. The HMDP allowed us to map Quantitative Trait Loci (eQTLs) with high resolution and power, identifying both local and trans acting variants—or, variants that affect a molecule from within and from outside, respectively.

Groups of genes are associated with multiple related phenotypes in HMDP, although not necessarily enriched for GO ontology or specific pathways. For more information, see our paper.

We report numerous novel transcripts supported by proteomic analyses, as well as novel non-coding RNAs. High resolution genetic mapping of transcript levels in HMDP reveals both local and trans expression eQTLs, identifying two trans eQTL ’hotspots’ associated with expression of hundreds of genes. We also report thousands of alternative splicing events regulated by genetic variants. We further showed that the genes associated with trans eQTL hotspots correlate to physiological phenotypes, such as HDL and triglyceride levels. This discovery provides insight into the mechanism behind correlation of these genotypes with complex traits.

Our data capture the various non-neuronal cell types, such as microglia or astrocytes, which are often overlooked in the mostly neuron focused studies of the hypothalamus. These cells are important mediators of hypothalamic inflammation and other processes induced by a high fat diet. Regulation of gene expression in these cell types impacts every aspect of metabolism, and our data provide a robust framework recapitulating transcriptional processes affecting multiple cell populations. Our approach is thus complementary to on-going cell type-specific transcriptomic efforts.

For more information, see our paper, which is available for download through eLife:

The full citation to our paper is: 

Hasin-Brumshtein, Yehudit; Hormozdiari, Farhad ; Martin, Lisa ; van Nas, Atila ; Eskin, Eleazar ; Lusis, Aldons J; Drake, Thomas A

Allele-specific expression and eQTL analysis in mouse adipose tissue. Journal Article

In: BMC Genomics, 15 (1), pp. 471, 2014, ISSN: 1471-2164.

Abstract | Links | BibTeX

See our blog post on a recent paper reviewing the HMDP data set:

Characterization of Expression Quantitative Trait Loci in Pedigrees from Colombia and Costa Rica Ascertained for Bipolar Disorder

Variants regulating gene expression (expression quantitative trait loci, eQTL) are at a high frequency among SNPs associated with complex traits. Genome-wide characterization of gene expression is an important tool in genetic mapping studies of complex disorders, including many psychiatric disorders. Further, implicating eQTL to specific tissue types is key to understanding functional variation in disease development. Our group, in collaboration with Chiara Sabatti (Statistics, Stanford) and Nelson B. Freimer (David Geffen School of Medicine, UCLA), developed a novel approach for analyzing eQTL and applied the method to a dataset from a bipolar disorder study.

Current approaches to implicating eQTL specific to tissues lack sufficient power in large-scale studies of human brain related traits, such as bipolar disorder. Together with the University of California San Francisco, Universidad de Costa Rica, Universidad de Antioquia, Medellín, Colombia, and Tel Aviv University, our group adopted a novel approach to assess the heritability and genetic regulation of gene expression related to bipolar disorder in populations from Costa Rica and Colombia.

This project examines 786 genotyped subjects originally recruited in a study of bipolar disorder, all related within 26 extended families. While the subjects in this study were originally recruited as part of an investigation for severe bipolar disorder (BP1), we found no relationship between the observed gene expression data and BP1. Instead, we use this unique Latin American population to explore the architecture of genetic regulation. Specifically, we estimate heritability, evaluate the relative importance of local vs. distal genomic variation, identify variants with regulatory effects, and analyze the role of multiple associated SNPs in the same region.

Our group adopted a novel hierarchical testing procedure that leads to the analysis of eQTL data in a stage-wise manner with increasing levels of detail. This design allows us to compare estimates of the heritability of gene expression obtained using both traditional and genotype-based methods. First, we apply a multiscale testing strategy to identify SNPs that have regulatory effects (eSNPs) on BP1. Second, we investigate which specific probes are influenced by these eSNPs. This hierarchical testing procedure effectively controls error rates and leverages the heterogeneity across genetic variants to preserve computational power.

We use this approach to measure gene expression in lymphoblastoid cell lines (LCLs) in subjects from extended families, segregating for BP1. Our results suggest that variation in expression values is heritable and that, at least in samples including related individuals, relying on theoretical kinship coefficients or on realized genotype correlation for estimation of heritability leads to similar results.

Expression heritability and proportion of genetic variance due to local effects. For more information, see our paper. For more information, see our paper.

Variance decomposition approaches suggest that on average 30% of the genetic variance is due to local regulation. In the majority of probes under local regulation in our sample, more than one typed SNP is required to account for expression variation. This finding can be interpreted as the result of heterogeneity, but also could reflect un-typed causal variants that are tracked by more than one typed SNP.

The knowledge we acquired by studying the genetic regulatory network within these pedigrees, instead, can be used to inform our mapping studies: eSNPs might receive a higher prior probability of association, or be assigned a larger portion of the allowed global error rate when using a weighted approach to testing. We will report elsewhere on the results of these investigations.

For more information, see our paper, which is available for download through PLoS Genetics:

The full citation to our paper is: 

Peterson, C.B., Jasinska, A.J., Gao, F., Zelaya, I., Teshiba, T.M., Bearden, C.E., Cantor, R.M., Reus, V.I., Macaya, G., López-Jaramillo, C. and Bogomolov, M., 2016. Characterization of Expression Quantitative Trait Loci in Pedigrees from Colombia and Costa Rica Ascertained for Bipolar Disorder. PLoS Genet, 12(5), p.e1006046.