NEW YORK (GenomeWeb) – Structural variants make up a greater portion of expression-altering variants than previously thought, according to a new study.
Using deep whole-genome sequencing data, researchers from Washington University School of Medicine and elsewhere mapped cis expression quantitative trait loci in more than a dozen tissues to examine the contributions of SVs, single-nucleotide variants, and insertions and deletions. As they reported today in Nature Genetics, the researchers estimated that SVs are causal in some 5 percent of eQTLs and generally have larger effect sizes than SNVs and indels.
"Our results show that SVs comprise a substantial and disproportionately large fraction of expression-altering genetic variants, a large portion of which have been untested in typical association studies," WashU's Ira Hall and his colleagues wrote in their paper.
The WashU team used multi-tissue RNA-seq expression data from the Genotype-Tissue Expression project to perform eQTL mapping based on deep whole-genome sequencing data from nearly 150 people. Through this, they uncovered 23,602 SVs and 24,884 cis eQTLs in 13 human tissues.
To overcome the limitations of previous SV-eQTL mapping attempts, the researchers gauged SV genotypes directly from sequencing data reads and relied on expression profiles from human tissues, not derived cell lines.
Through joint eQTL mapping, the researchers gauged the contributions of structural variants to variations in expression. SVs, they reported, were the lead marker at 3.5 percent of eQTLs, while SNVs were at 85.5 percent and indels at 10.9 percent of eQTLs. In whole blood in particular, they found that SVs contributed to 2.2 percent of eQTLs — some four-fold higher than a previous estimate made by the 1000 Genomes Project, the researchers reported.
Using fine mapping, the researchers uncovered 789 possibly causal SV-eQTLs. They estimated that for between 3.5 percent and 6.8 percent of eQTLs — depending on the causal-variant inference method used — structural variants were predicted to be the causal variant.
Hall and his colleagues also estimated that structural variants were likely to have a greater influence on expression than SNVs and indels. In particular, they reported that an SV was 28 to 54 times more likely to modulate expression and that an SV has a median 1.3-fold larger effect size on gene expression than SNVs and indels.
This indicted to the researchers that SVs have a key and disproportionally large role in genetically regulated gene expression.
Most of the predicted causal non-coding structural variants were enriched for overlaps with regulatory regions, the researchers added. Some 88 percent of the 789 putative causal SVs were enriched at enhancers and other regulatory elements.
In addition, 52 of these putative causal SVs were in linkage disequilibrium with loci uncovered through genome-wide association studies. According to the researchers, this set of loci includes a number of promising candidates, including a 294-base pair deletion that's associated with decreased expression — possibly by disrupting an enhancer — of the DAB2IP gene in thyroid tissue, which has itself been linked to a risk allele for abdominal aortic aneurysm.
Additionally, they noted that rare variants, especially SVs, were enriched in regions near gene expression outliers, which they said suggests that rare SVs are a common cause of aberrantly expressed genes.
"These results suggest that comprehensive WGS-based SV analyses will increase the power of common- and rare-variant association studies," Hall and his colleagues noted.