Efficient methods for gene mapping of complex diseases.

Abstract

Genome-wide association (GWA) studies are a promising strategy for identifying common genetic variants that predispose to human disease. Because of the high cost of genotyping hundreds of thousands of markers on thousands of subjects, GWA studies will often follow a staged design in which a proportion of the available sample is genotyped on all markers in stage 1, and a proportion of these markers are followed up on the remaining samples in stage 2. The standard strategy for analyzing such two-stage data views stage 2 as a replication study, and focuses on findings that reach statistical significance when stage 2 data are considered alone. In this dissertation I demonstrate that the alternative strategy of jointly analyzing the data from both stages almost always results in increased power to detect genetic association, despite the need to use more stringent significance levels. I also demonstrate how to design two-stage GWA studies that balance the need for high power, low false-positive rate, and reasonable cost. I show how the cost and design of these studies are influenced by the ratio of stage 2 to stage 1 per genotype cost, and how genotyping costs can be controlled by modestly decreasing the power achieved while maintaining the false-positive rate or by relaxing the false-positive rate while maintaining power. One obstacle to identifying variants predisposing to human disease is population substructure. When unaccounted for, it can lead to excess false positives and loss of power in both linkage and association studies. Methods exist which can identify substructure by modeling different allele frequency distributions for a fixed number of founder populations, or through inflation of the non-centrality parameter of the association test statistic. However, it is unclear how to identify genetically homogenous subsets of individuals or families when little genotype data is unavailable, which may be the case during the early phase of a linkage study. I develop a simple algorithm that accomplishes this by transforming self-reported ethnicity and pedigree structure into a family ancestry classification. I apply this algorithm to the self-reported ethnicity information of 159 families from a schizophrenia linkage study. I compare the estimates of population membership to those obtained using a <italic> structure</italic> analysis of 378 microsatellite markers and find excellent concordance between the two.