Genetic epidemiologic design focusing on African Americans or Hispanics is particularly challenging because both groups have experienced recent admixture. Indeed, in the presence of cryptic population structure, case-control association designs can produce false-positive association findings. However, case-control designs have unique advantages that family-based designs cannot replace. Therefore, it is crucial to develop approaches that analyze case-control studies and are immune to the perils of population stratification. At the same time, recent admixture creates strong linkage disequilibrium, providing exciting opportunities for novel disease-mapping approaches. The long-term goal of this research is to develop novel quantitative methods which enable researchers to identify the factors that influence disease risk in admixed or stratified populations. This project aims to develop new methods that improve the robustness and efficiency of casecontrol association studies, with applications to African Americans and Mexican Americans. Bootstrap- and resampling-based procedures will be developed to infer important aspects of an individual's ancestry, using genotype data at linked and unlinked genetic markers (Aim 1).
Aim 2 strives to shed light on a long-standing controversy regarding the scope of confounding that is likely to occur in practice.
Aim 3 develops an adjustment-based approach which incorporates the estimated individual admixture in a regression model and thereby explicitly controls for confounding due to population stratification. Finally, Aim 4 proposes a linkage-disequilibrium mapping approach. The strengths and limitations of the adjustment approach (Aim 3) and the linkage disequilibrium mapping approach (Aim 4) will be assessed analytically and through simulations. Results from this research will enable investigators to better identify and control for bias due to admixture in population-based case-control studies, and to design new valid and more powerful studies. Genotypic data from various multiethnic studies will be used together with extensive simulation experiments to test and refine the methodology for real-world applications.
Showing the most recent 10 out of 33 publications
