Genetic surveys in human populations often involve a mixture of subpopulations. Recognition of the presence of mixtures and identification of mixture components is essential in understanding the maintenance of genetic variation, and for reconstructing evolutionary history. This project aims at a quantitative assessment of the effect of admixture on genetic variation in natural populations. Two types of mixtures will be studied: (a) mixture due to amalgamation of subpopulations; and (b) admixture due to gene flow among source populations. In both instances, we will develop statistical methods to identify the source of mixtures, and estimate the mixing components. Since the presence of mixture affects the observed allele frequency distribution, adjustment for the effects of mixture components will provide reliable estimates of locus-specific mutation rates (v's) and effective population size (e's) from multi-locus allele frequency data. The utility of admixed populations for genetic- epidemiological studies of complex diseases will be examined by the dynamics of allelic association among loci and its relationship with the history of admixture. The maximum likelihood principle will be used to develop analytical methods for such analyses, and numerical algorithms will be provided to encompass the possibility of missing data and unknown source populations. The theory will be applied to allozyme data from the Amerindians, and the three major ethnic groups, to test its validity, and to provide empirical observations on the effect of mixtures on genetic variation. Genetic evidence of admixture will be studied in the Siddis of India, and the Sinhalese of Sri Lanka, whose parental populations are not precisely known. Extensive computer simulations and re-analysis of published data from Asian Macaques and Drosophila will be used to test the theories and to study the statistical properties of the estimators. The estimation of locus-specific mutation rates will contribute to an understanding of the variation of mutability at different loci, or at DNA segments in different parts of the human genome. Simultaneous estimates of v's and Ne's will allow effective comparisons of the rate of spontaneous mutations in different organisms. The relationship between admixture levels and the multi-locus association of alleles will aid (a) in understanding the genetic etiology of complex diseases, and (b) in developing efficient survey strategies for mapping genes underlying such diseases.
Showing the most recent 10 out of 99 publications
