It has become widely recognized that most complex diseases result from many genes, with each gene only having a modest effect on the disease. Linkage analysis, a very effective method at mapping rare disease for which single gene is sufficient to cause disease, is less successful in localizing complex disease genes. It is believed that direct association approach is more powerful for identify common variants conferring modest risk. For a whole-genome scan, however, direct association is currently impractical, because it requires the genotypes at hundreds of thousands or millions of markers. Admixture mapping, an association-based approach using the information generated by recent population admixture, offers a promising but as yet untested methods for performing a whole-genome scan. The key advantage of admixture mapping is that while it is based on directly associating sections of the genome with disease, it only needs about 1% of markers for direct association studies. Although the idea of admixture mapping is simple, its practical application has awaited the development of statistical methods. This project will develop novel statistical methods for admixture mapping in whole-genome scan and further in candidate region studies using the sample from an admixture population, especially African-American population. First, a new method to estimate the ancestral status and allele frequencies in the founding populations will be developed. Based on the estimation method, statistical tests to test for linkage in genome scan and the tests to test for linkage and for association separately in candidate region studies will be proposed. In this project, it is also proposed to explore the pattern of LD caused by the recent admixture of two founding populations, and to evaluate the precision and the power of the methods by using extensive simulation studies.
