Model-free lod score methods are used extensively to map genes underlying complex disorders using samples of affected sib pairs. Since 1990, extensions to the basic one-disease locus model have been proposed that allow separate analysis or inclusion of other types of relative pairs, discordant relative pairs, covariates (i.e., gene by environment interaction), or additional disease loci have been proposed. Recently, I proposed the conditional logistic model that reparameterizes the problem in terms of relative risks and thus provides a unified framework that allows all these features to be incorporated. At present, the simple two-parameter model, which automatically includes all types of affected relative pairs, has been incorporated into the computer program SIBPAL2 in S.A.G.E. and is available through the S.A.G.E. web site as a beta version of the program. The goal of this proposal is further development, study, implementation and dissemination of the conditional logistic method. New methodology to be developed includes 1) an optimal one-parameter model that extends to include covariates and thus increases the power of tests of gene-by-environment interaction, 2) information weighting of pairs to increase the power of tests of linkage, 3) joint disequilibrium and linkage modeling, and 4) extension of the methods of X-linkage. In addition, large and small sample properties of test statistics will be studied. These extensions, as well as extensions previously proposed (Olson, 1999) will be incorporated into the SIBPAL2 program. At the end of the project period, a comprehensive set of tools for model-free affected relative pair analysis will be available.
