Constructing genetic and physical maps for all the human chromosomes are fundamental aims of the Human Genome Project, and the promise to map and identify genes for human diseases is a principal reason why the Genome Project has widespread support. Efficient statistical methods help to insure that these goals are met in a timely, cost-effective manner. In this proposal, I address a set of statistical design and analysis issues that arise in radiation hybrid map construction, genetic map construction, and the mapping of human disease genes. For radiation hybrid mapping, I will consider (1) questions of study design, including sample size requirements and accuracy of locus ordering; (2) methods of framework map construction; (3) extensions to standard models to allow for (a) combining data from two or more hybrid panels, (b) permit exclusion of hybrids that appear not to contain human material, and (c) allow for direct marker selection in hybrid construction; and (4) methods to identify typing errors and to assess the magnitude of their effect on distance estimates and accuracy of ordering. For genetic map construction and the mapping of disease genes, I will consider (1) methods for error detection in gene mapping studies; (2) decision rules for changing from two-point to multipoint linkage analysis when mapping a disease gene; (3) Bayesian methods for determining the posterior probabilities of autosomal and X-linkage; and (4) two-locus methods for,mapping complex genetic diseases. In addition, I will continue to develop, support, and distribute my computer program SIMLINK, the standard tool for evaluating the power of a proposed linkage study. Finally, I will continue to be opportunistic in identifying and addressing important statistical modeling and analysis problems that arise that are related to the general goals of this project.
Showing the most recent 10 out of 67 publications
