Student Theses and Dissertations

Date of Award

2010

Document Type

Thesis

RU Laboratory

Breslow Laboratory

Keywords

population isolates, genetic mapping, genotype mapping, haplotypic mapping, Kosrae, Ashkenazi Jews

Abstract

The potential benefits of using population isolates in genetic mapping due to reduced genetic and environmental heterogeneity are offset by the challenges posed by these populations for traditional association methods. Population isolates often contain large amounts of direct and cryptic relatedness that confound baseline assumptions of independence among genotypes and phenotypes and require specialized approaches to account for this sample structure. We examined three such approaches for association testing: (i) scoring allele transmission to offspring within families (ii) incorporating a permutation-based association score between families into the test statistic and finally (iii) incorporation of a kinship matrix to capture the relatedness among all individuals into a mixed model to test for association. The mixed model approach had 88% power to rank the true SNP as among the top 10 genome-wide top 10 with 56% achieving genomewide significance, a >80% improvement over other methods. We then used the mixed model method for genome scans relating to metabolic traits and electrocardiographic measures in 2,906 related individuals from the Island of Kosrae, Federated States of Micronesia, who were previously genotyped for over 330,000 SNPs. We re-analyzed data for 17 published and 8 previously unpublished metabolic and electrocardiographic traits. We replicate seven genome-wide significant associations with known loci of plasma cholesterol, high density lipoprotein, low density lipoprotein, triglycerides, thyroid stimulating hormone and C-reactive protein, with only one detected in the previous analysis of the same traits. We further report novel associations for height (rs17629022, p<2.1E-8), homocysteine (rs7481043, p<1.3E-8) and uric acid (rs2186571, p<1.8E-34), the latter two near relevant candidate genes. We demonstrated the increased power of mixed-models for handling hidden and direct relatedness in isolated cohorts and discovered three novel associations with height, homocysteine and urate levels. Our experiences in association testing in an isolated population can serve as a model for other studies of similar cohorts.

Comments

A thesis presented to the faculty of The Rockefeller University in partial fulfillment of the requirements for the degree of Doctor of Philosophy.

Permanent URL

http://hdl.handle.net/10209/342

Included in

Life Sciences Commons

Share

COinS