Risk alleles for environment-depended diseases are selectively neutral

Reviewing thousands of genome wide associate studies (GWAS) to identify genetic variants in single nucleotide polymorphisms (SNPs), investigators at Dartmouth’s Norris Cotton Cancer Center found that some alleles (one of a pair of genes located on a specific chromosome) are more frequently risk-associated with disease than protective. The research, conducted by Ivan P. Gorlov, PhD, Olga Y. Gorlova, PhD, and Christopher I. Amos, PhD of the newly formed Department of Biomedical Data Science, was shared in “Allelic Spectra of Risk SNPs Are Different for Environment/Lifestyle Dependent versus Independent Diseases,” and published in PLOS Genetics.

“SNPs usually have two alleles: minor (with population frequency <50%) and major (with population frequency >50%),” said Gorlov, first author of the paper. “With respect to disease, alleles can be categorized into being risk or protective ones. We researched if the proportion of the risk-associated alleles was higher among minor alleles and found selection against risk alleles is drives their frequencies down. Because of this, risk alleles are overrepresented among minor alleles.”

To further their research, the team used an environment/lifestyle index (ELI) to divide diseases into two groups: environment/lifestyle dependent (E/LS) and E/LS independent. The first group includes such diseases as type 2 diabetes and cardiovascular disease, whose incidence is linked to recent changes in human environment and lifestyle. The second group included diseases with no or little evidence that their incidence depends on environment and lifestyle. They found that the risk allele distribution for E/LS dependent diseases follows a neutral model while there is clear evidence of negative selection for E/LS independent disease.

“We hypothesized that changes in the environment and lifestyle reassign functional significance of existing polymorphisms, turning some of them into risk-associated and others into disease protective,” Gorlov said. “Because the absolute majority of existing polymorphisms are selectively neutral, the frequency distribution for disease associated SNPs follows a neutral model. It takes several generations for selection to reduce frequencies of risk-associated alleles which will result in their overrepresentation among minor variants as we observe for E/LS independent diseases.”

Human diseases with the largest public burden are E/LS dependent diseases, and the team found that allelic distributions for them follow that for an average human SNP which is neutral. This will allow them to estimate percentage of heritability that can be explained by all disease-associated SNP. Estimating the total SNP contribution in disease heritability will determine if there are other significant sources of inherited risk.

The next step for Gorlov and the team is to apply a neutral model to estimate the total heritability of E/LS detected disease attributable to all disease-associated SNPS, including those that are not yet detected. These findings also will help investigators to understand the etiology of common complex diseases because we can anticipate more functionally significant variants to influence diseases with a stronger environmental component, while uncommon variants may be more important for diseases with a weak environmental contribution.

For more information, contact Kirk Cassels at (603) 653-6177.

Link to original article publication: http://cancer.dartmouth.edu/about_us/article/35107
August 25th, 2015