Novel Statistical Methods for Human Gene Mapping

人类基因图谱的新统计方法

• 批准号: 7146455
• 负责人: MICHAEL PHILIP EPSTEIN
• 金额: $ 30.6万
• 依托单位: EMORY UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-09-27 至 2011-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7146455
• 关键词: clinical research; genes; genetics; human; model

DESCRIPTION (provided by applicant): Many common human diseases originate in part from the complicated effects of multiple genetic variants found throughout the genome. Given the enormous impact of such diseases on public health, it is imperative to map relevant genetic variants to improve our understanding of the molecular basis of such diseases, as well as improve screening techniques for disease prevention. To this end, successful human gene mapping of complex diseases requires the development and application of powerful statistical methods that fully utilize the resources of the Human Genome Project. This grant proposes a set of such statistical methods that either address novel problems or improve existing solutions to problems in human gene mapping studies. These proposed methods are applicable to a variety of genetic studies as they address topics in linkage, linkage disequilibrium, and high-dimensional genetic analyses of complex diseases and disease-related quantitative traits. The methods can be partitioned into the two general groups: mixed-modeling procedures and case-control likelihood procedures. The mixed-modeling procedures considered include a general variance-component (VC) mapping framework for continuous and discrete trait data and a modified VC mapping framework that allows for haplotypes. Also considered is a novel linear-mixed-model framework that identifies a large combination of genetic variants that influence a quantitative trait using support-vector- machine regression. The case-control likelihood procedures considered are extensions of the approach of Epstein and Satten (2003) for haplotype inference on disease. Extensions considered include allowing for covariates and haplotype-covariate interactions, and also categorical disease outcomes. The proposed statistical methods in this grant have the potential to increase the power to identify genetic variants that influence complex diseases and disease-related quantitative traits. This project will evaluate the performance of these methods using simulations based on the study design and data from an existing gene mapping study of type 2 diabetes. Also, this grant will implement the proposed statistical methods in user- friendly, efficient software for public distribution. Finally, this project will be opportunistic in identifying and addressing unforseen statistical problems arising in human gene mapping studies.

描述（由申请人提供）：许多常见的人类疾病部分源于在整个基因组中发现的多种遗传变异的复杂影响。鉴于此类疾病对公共卫生的巨大影响，必须绘制相关遗传变异图谱，以提高我们对此类疾病分子基础的理解，并改进疾病预防的筛查技术。为此，成功绘制复杂疾病的人类基因图谱需要开发和应用充分利用人类基因组计划资源的强大统计方法。这项资助提出了一套这样的统计方法，要么解决新的问题，或改善现有的解决方案，在人类基因图谱研究的问题。这些建议的方法适用于各种遗传研究，因为它们解决了连锁，连锁不平衡和高维复杂疾病和疾病相关的数量性状的遗传分析的主题。这些方法可以分为两大类：混合建模方法和病例对照似然方法。考虑的混合建模程序包括一个通用的方差分量（VC）映射框架的连续和离散性状数据和修改后的VC映射框架，允许单倍型。还考虑了一种新的线性混合模型框架，该框架使用支持向量机回归识别影响数量性状的遗传变异的大组合。所考虑的病例对照似然程序是Epstein和Satten（2003）对疾病单倍型推断方法的扩展。考虑的扩展包括允许协变量和单倍型-协变量相互作用，以及分类疾病结果。这项研究中提出的统计方法有可能提高识别影响复杂疾病和疾病相关数量性状的遗传变异的能力。该项目将使用基于研究设计和现有2型糖尿病基因图谱研究数据的模拟来评估这些方法的性能。此外，这笔赠款将实施拟议的统计方法，在用户友好，高效的软件，供公众分发.最后，这个项目将在识别和解决人类基因图谱研究中出现的不可预见的统计问题方面具有机会性。