COLD SPRING HARBOR, NY (GenomeWeb News) – Synthetic associations between rare, disease-causing variants and common variants probably explain relatively few of the signals detected in genome-wide association studies of common disease, according to Wellcome Trust Sanger Institute human genetics researcher Jeffrey Barrett.
"The preponderance of evidence suggests they're very unlikely to underlie many GWAS hits," Barrett said.
Speaking at the Biology of Genomes meeting here yesterday, Barrett argued that such synthetic associations are possible but probably fairly uncommon based on linkage data, local re-sequencing, pathway analyses, and trans-continental replication studies of disease-associated variants.
The analysis came in response to papers by David Goldstein and colleagues highlighting the potential role of rare variants in GWAS. In a PLoS Biology paper this January, for example, Goldstein and his co-workers used GWAS and simulation studies to show that rare variants can produce GWAS hits through synthetic associations with common variants.
In another paper appearing in the American Journal of Human Genetics last month, the team estimated the effect size and proportion of variance explained by rare variants, arguing that "thinking beyond common variants is essential in interpreting GWAS signals and identifying causal variants."
Although he said synthetic associations are "plausible," Barrett said there is currently little evidence supporting the notion that these interactions underlie many or most signals detected in GWAS of common disease.
"There's an excellent, known example of synthetic association," Barrett said, pointing to Crohn's disease — a condition in which rare, disease-causing mutations in a gene called NOD2 interact with neighboring common variants to produce signals that can be detected during GWAS.
But for several other diseases studied so far, linkage analyses, re-sequencing data, pathway analyses, or other types of genomic information don't appear to support a role for rare variants in GWAS hits, Barrett argued, providing examples from type 1 diabetes and other conditions.
In the case of type 1 diabetes, for instance, he noted that dozens of genes have been associated with disease. But while rare variants have been detected in diabetes, re-sequencing studies suggest these variants are independent of GWAS hits, Barrett said.
And in contrast to the idea that distinct, more penetrant rare disease-associated variants have cropped up in different populations, Barrett highlighted data suggesting many GWAS associations seem to replicate and show similar directional effects across diverse populations.
"We believe the consideration of these wider lines of evidence support the conclusion that synthetic associations explain very few GWAS signals," Barrett and his co-authors wrote in the talk's abstract.