| Summary: | The concerted effort to characterize the contribution of inherited variation to human health and disease through genome wide association studies (GWAS) has promised to increase our understanding of the diverse molecular pathways underlying specific human traits, as well as yield clinically actionable findings that would prove useful in estimating disease risk and designing personalized therapeutic approaches. However, it has become evident in recent years that further work is needed in order to harness the biological and translational potential of GWAS findings.
One approach to glean further insight from GWAS stems from the seminal discovery that trait-associated SNPs are enriched for expression quantitative trait loci (eQTL), highlighting an important mechanism by which SNPs influence traits, i.e. through modulation of gene transcript levels. Similarly, valuable insight into biological mechanisms can be gained through expanding eQTL studies to include genetic regulation of non-coding RNA levels, whose inter-individual variation, like gene expression, has been shown to impact on complex diseases.
Here we identified a large collection of distant miRQTL (391) and replicated with 26% p < 0.05 and 98.5% allelic concordance (67 of 68 SNPs) in an independent cohort. Analysis of genomic properties of replicated miRQTL reveal strong enrichment for mapping within ENCODE-annotated functional elements. In-silico analysis using replicated distant miRQTL reveal local mRNA expression quantitative trait loci (eQTL) putatively regulating microRNA abundance. Mediation analysis and association testing between microRNA and mRNA confirms HEXIM1 as a putative regulator of hsa-mir-185-5p levels. These results highlight a potential novel mechanism of long-range regulation of microRNA abundance, providing valuable insight into the biology underlying complex traits.
A second approach to transform GWAS findings into clinically relevant tools is based on developing and applying statistical prediction methods to existing GWAS information. These approaches have the potential to translate genetic data into clinically relevant predictions of risk. It is these high-confidence predictions of complex traits such as disease risk or drug response that will fulfill the goal of personalized medicine.
Therefore we proposed a novel systems approach to complex trait prediction, which leverages and integrates similarity in genetic, transcriptomic or other omics-level data. Using seven disease datasets from the Wellcome Trust Case Control Consortium (WTCCC), we show that OmicKriging has important translational potential. In addition, we built a statistical-learning machine, which integrates large-scale whole-genome data to predict bevacizumab-induced hypertension in cancer patients. We found that incorporating primary genetic as well as clinical trial data into our model significantly improves prediction and therefore should motivate the use of such large-scale whole-genome predictors in a clinical setting. Taken together, these approaches utilize novel methodology as well as publicly available datasets to yield valuable mechanistic insight into genetic regulation of complex traits, and provide genetic tools for clinical implementation.
|
|---|
