Year: 2002
Type: dissertation
Abstract: Complex diseases are thought to be caused by both environmental and genetic factors. Single nucleotide polymorphisms (SNPs) are currently being explored for use as genetic markers in association studies of complex diseases. SNPs are abundant in the genome, leading to much finer scans of candidate regions. Since multiple SNPs in a region are likely in linkage disequilibrium, it has been suggested that methods which use the information at several SNPs at a time, along the haplotype, will be better for finding disease-predisposing genes through association studies. For certain genotypes, it is not possible to determine the haplotype phase so missing data methods have been used to infer the haplotype or the haplotype frequencies in a sample. The inferred haplotypes or frequencies are then used in association analyses. A chi square test, either on the inferred haplotype counts or the estimated haplotype frequencies, is commonly used but does not account for environmental factors. Currently, many association analyses which adjust for environmental cofactors use imputed haplotypes but do not account for this uncertainty. In this project, the expectation maximization (EM) algorithm is used to handle the missing haplotype information in a logistic regression with haplotypes as the genetic covariates and other fully observed environmental covariates. The haplotypes are not imputed, instead the EM algorithm is used to obtain maximum likelihood estimates of the regression coefficients directly. The variance-covariance matrix is derived using Louis' formula for the observed information (1982) and the method is applied to a simulated cohort study dataset. This method, which allows for both genetic and environmental factors and does not assume that the haplotypes are fully observed, is shown to find the genetic signal better than an analysis which does not include environmental factors.
Source: Summit (Simon Fraser University)
Author: Kelly M. Burkett
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2138436009