INFERENCE AND ROBUST METHODS FOR LINKAGE MAPPING

描述（改编自研究者摘要）：近年来，

DESCRIPTION (Adapted from the Investigator's Abstract): In recent years, great methodological and computational advances have been made in genetic linkage mapping. However, several important tissues have received insufficient attention. The long-term objectives of this project are to develop new statistical and computational methods for discovering and localizing genes influencing complex traits and to improve the design of genetic linkage studies. All of the methods apply directly to experimental crosses and human studies involving relative pairs, and so are of immediate biomedical importance. In addition, the insight gathered from these investigations may enable extensions to more genetic epidemiological designs including the parametric analysis of multiplex pedigrees. The specific aims are: I. to develop general and flexible methods for constructing confidence intervals for gene locations in genetic linkage studies. A method proposed here is surprisingly simple and straightforward , and can be implemented immediately by researchers using existing software for multipoint mapping. II. to unify several linkage mapping approaches, to obtain an improved understanding of the connections among various genetic epidemiological designs. In addition, the unification appears to enable simple and powerful investigation of efficiency and precision of various linkage designs. III. to develop extremely general non-parametric approaches to linkage analysis of QTLS using empirical distribution functions of phenotypes conditioned of IBD status, and to develop adaptive pseudo-likelihood methods for efficient inference for the gene location. These aims will be achieved using both analytic and computational approaches, and any computer algorithms developed for this purpose will be distributed for public use. In addition, many of the methods proposed can be easily implemented by other investigators using existing software. The robustness and wide applicability of the methods proposed will be examined using a combination of simulation and analysis of existing linkage datasets.