MODELS FOR THE GENETIC EPIDEMIOLOGY OF CHRONIC DISEASE

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): This is a competing continuation application for years 21-25 of a project to develop and implement models and methodology for the likelihood analysis of pedigree data and to distribute the resultant software to the research community. Each aspect of the proposal involves PAP (Pedigree Analysis Package). Parametric genetic analysis, the primary strength of PAP, should play a role in the multi-pronged approach needed to tackle the difficult task of characterizing and localizing genes for complex diseases, which result from complicated interactions of multiple genes and environmental factors. Non-parametric linkage analysis will also become available in PAP upon completion of extensions proposed in this application. Although there exist a number of other software packages for the analysis of family data, none matches the versatility of PAP. The extension of PAP more efficiently produces software for diverse genetic analysis than does the development of new software for that purpose. In addition, PAP has a track record for having been wellsupported over the years. During the previous funding period, a graphical user interface in Java was developed to facilitate the use of PAP. During the remaining year of funding the conversion of the source code will be completed, thereby making PAP very portable and convenient to install while maintaining all the analysis versatility. Nevertheless, PAP remains impractical for multipoint linkage analysis, because of the memory and computer time requirements. Therefore, this application proposes to remedy that shortcoming by implementing Markov chain Monte Carlo (MCMC) methods for the analysis of multi-locus marker data. MCMC methods sample, rather than exhaustively enumerate, the possible combinations of genotypes within a pedigree, therefore effecting an enormous saving of computer time. With the availability of MCMCproduced probabilities, multipoint linkage analysis will become feasible for either variance components models or using any of the extensive selection of major gene models already available in PAP. Nevertheless, the current application also proposes to develop new genetic models of disease risk, including a mode-of-inheritance free method. In addition, this application proposes to implement changes to increase computational speed, allow data exploration, and facilitate data handling and analysis, while continuing to distribute and support PAP.