Multipoint methods of linkage analysis provide increased power for detecting linkage over two-point methods and facilitate construction of detailed human genetic maps. In addition, increasing attention to diseases and traits of complex etiology has led to renewed interest in linkage methods that do not rely on full specification of the trait genetic model, such as affected relative pair methods and the Haseman-Elston sibpair method for mapping quantitative trait loci. The proposed research will develop in detail the necessary theory for the robust mapping of loci that underlie complex human diseases and quantitative traits. The methods will be based on relative pair analysis and will be designed to utilize information from more than one marker and from large pedigrees by combining interval mapping methods with quasilikelihood and pseudolikelihood techniques for combining information from correlated relative pairs. The methods thus will provide more powerful tests for linkage and estimates of gene location that do not rely on (complete) specification of the genetic model for the trait. Alternative methods and means of implementation of the methods will also be developed and investigated, including randomization tests for determining p-values, two/multi-stage sampling strategies, variance-components approaches, and the use of exact covariance structures. Competing methodologies will be evaluated on the basis of power, validity and robustness, using both real and simulated data, to determine the most effective strategies for studying genetic linkage for complex diseases. The information obtained from this research program will be used to direct development of genetic statistical software in conjunction with the S.A.G.E. project.
