Using SNPs to study complex genetic disease : a population and evolutionary genetics perspective

Abstract

Associating genes to complex genetic diseases has proven more difficult than anticipated. Many genetic variants contribute small effects towards an individual's disease predisposition in polygenic diseases, and this biological complexity requires new methods of analysis. This thesis studies genetic variation from a population genetics and evolutionary perspective, to prioritise single nucleotide polymorphisms (SNPs) and to understand the patterns of linkage disequilibrium globally, in order to facilitate disease association studies.

Genotyping technologies are an integral part of the disease gene-mapping field, and thus the first work investigates whether Dynamic Allele-Specific Hybridization (DASH) is able to score insertion-deletion (indel) polymorphism. Indel markers in coding regions can produce frameshift changes, potentially altering protein function. Design criteria for indel genotyping was established, after which indel markers in candidate genes for Alzheimer's disease (AD) were used in an association study to test for correlation of the variant to disease status. While no associations were found between AD and the markers, we confirmed that DASH can be recommended for indel genotyping.

The second project investigated stop-codon polymorphism in the genome. The difficulties of associating genetic markers to disease in linkage disequilibrium (LD) mapping studies encouraged us to prioritise SNPs that had the greatest chance of being the susceptibility variant for disease. Stopcodons prematurely truncate proteins and are expected to have a phenotypic function. In our study of 50 stop-codon polymorphisms, a mere 2 were shown to be real after initial validation studies, suggesting that functional variation is under selective constraints and have high false prediction rates in databases. This is important to consider for study design.

The third investigation of this thesis looks at linkage disequilibrium and haplotype patterns in 16 populations. A large-scale international effort was begun to map human variation in four populations, in order to facilitate disease studies. Our study investigates the effects of different demographic histories of populations on common variants to determine if the 'HapMap' will be informative in all populations. We suggest that there are both similarities and differences between global populations, and extrapolating data from one population does not always capture a true picture of diversity in other populations.

The fourth project investigates the global allele frequencies, as estimated from 16 populations, of non-synonymous SNPs (nsSNPs) and their predicted functionality. We began by validating 1300 nsSNPs in African American and European American samples, creating a resource of 500 SNPs for disease studies. 200 of these SNPs were genotyped in pooled populations and we confirmed previous studies showing that damaging variants are under negative selection in the genome. SNPs that were damaging, but showing large global variation in allele frequencies, were most often in genes associated with disease. This suggests that continued efforts looking for the effects of selection in the genome will be rewarding for diseasemapping studies.