Discovery and characterization of genetic variants associated with extreme longevity

Abstract

Over the last decade, there have been multiple genome-wide association studies (GWASs) of human extreme longevity (EL). However, only a limited number of genetic variants have been identified as significant, and only few of these variants have been replicated in independent studies. There are two possible reasons for this limitation. First, genetic variants might have a varying effect on EL in different populations, and GWAS applied to a dataset as a whole may not pinpoint such differences. Second, EL is a very rare trait in a population, and rare and uncommon variants might be important factors in explaining its heritability but GWASs have focused on the analyses of variants that are relatively common in the population. In this dissertation, I present three projects that address these issues. First, I propose PopCluster: an algorithm that automatically discovers subsets of individuals in which the genetic effects of a variant are statistically different. PopCluster provides a simple framework to directly analyze genotype data without prior knowledge of subjects ethnicities. Second, I investigate ethnic-specific effects of APOE alleles on EL in Europeans. APOE is a well-studied gene with multiple effects on aging and longevity. The gene has 3 alleles: e2, e3 and e4, whose frequencies vary by ethnicity. I identify several ethnically different clusters in which the effect of the e2 and e4 alleles on EL changes substantially. Furthermore, I investigate the interaction of APOE alleles with the country of residence. Results of this analysis suggest possible interaction of this gene with dietary habits or other environmental factors. For the third project, I perform a GWAS of rare variants and EL in a case-control dataset with median age of cases 104 years old. I analyze 4.5 million high-imputation quality rare SNPs imputed with HRC panel with minor allele frequency < 0.05. The analysis replicates all previous genome-wide level significant SNPs and identifies a few more potential targets. Additionally, I use serum protein data available for a subset of subjects and find significant pQTLs which have potential functional role. Based on these analyses, both genetic and environmental factors appear to be important factors for EL.