While genome-wide association studies (GWAS) have identified tens of thousands of common genetic variants associated with complex traits, understanding the underlying mechanisms in terms of gene function is not straightforward. Eleonora Porcu, SIB Member and Postdoc at the Center for Integrative Genomics of the University of Lausanne under the supervision of Zoltán Kutalik and Alexandre Reymond, proposes a new method to estimate the causal effects of gene expression on human traits. The statistical approach, which combines information from GWAS and gene expression, is described in a paper published in Nature Communications.
Randomized controlled trials are the gold standard in inferring causality between a risk factor and a health outcome such as disease - however, they are not always ethical nor feasible. Enter Mendelian Randomization (MR), a statistical approach that is able to provide evidence, in observational studies, of putative causal relations between risk factors and disease, using genetic variants as natural experiments (Davies N M et al. BMJ, 2018). In the current study, MR was used to estimate the causal effect of gene expression on complex traits, i.e. traits showing a continuous range of variation and which are influenced by both environmental and genetic factors, such a height.
From association to causality: a statistical journey
Genome-wide association studies (GWAS) have led to the discovery of massive amounts of genetic variants associated with complex traits for instance, but the underlying functional mechanisms – and thus causation – often remain unclear.
Recently, it has been shown that a large fraction of these trait-associated variants is also associated with gene expression: they are known as expression Quantitative Trait Loci (eQTLs). This reflects their potential involvement in gene regulation, which represents a first step towards a better understanding of the mechanisms at stake and a key ingredient for the proposed Mendelian Randomization approach (see Box).
“Here we propose a method called TWMR – for Transcriptome-Wide Mendelian Randomization – which integrates summary statistics from GWAS and eQTLs studies in a Mendelian Randomization framework to estimate the causal effect of gene expression on several human phenotypes”.
Unlocking the hidden value of GWAS
The method has been applied to the largest publicly available GWAS summary statistics and combined them with eQTL data from GTEx (Genotype Tissue Expression Project) and the eQTLGen Consortium. It thus provided an atlas of putative functionally relevant genes for 43 complex human traits.
More specifically, it revealed 3,913 gene-trait associations, 27% of which were uncovered so far by conventional GWAS. Noteworthy among these novel links are the height- and educational attainment-associated genes RAB23 and BSCL2, known to carry mutations leading to, respectively, Carpenter syndrome and a mendelian form of encephalopathy.
Furthermore, the results show that causal genes can be highly tissue-specific, highlighting the importance to integrate data from the tissue of interest for the studied phenotype when trying to interpret GWAS results using gene expression as an intermediate phenotype.
“TWMR unlocks hidden value from published GWAS through higher power in detecting associations and has the means to unravel new biological mechanisms underlying complex and clinical traits,” says Porcu. “While the causal associations reported in this study are still putative, they provide a promising list of candidate genes for follow-up studies when much larger eQTLs sets become available.”
Application to other omics data
While in this study the authors used eQTLs data from gene-expression, their Mendelian Randomization approach can be applied to other “omics” (e.g. methylation, metabolomics, proteomics) data. Indeed, their method only requires summary statistics from GWAS and any kind of exposure partnered with linkage disequilibrium estimates.
“I believe that given the urgent need to understand how GWAS loci effectively contribute to the variation of human phenotypes, new methodologies such as Mendelian Randomization combining GWAS signals with relevant omics data will become an increasingly important approach for the identification of genes, pathways and networks that underlie human complex traits”, concludes Porcu.
Reference
Porcu E et al. Mendelian randomization integrating GWAS and eQTL data reveals genetic determinants of complex and clinical traits, Nature Communications 2019, doi: 10.1038/s41467-019-10936-0