Haplotype-Based Association Modeling for Whole-Genome Scan and Candidate Gene Studies

摘要提案编号：DMS-0504726机构：北卡罗莱纳州立大学提案题目：基于单倍型的全基因组扫描关联建模和候选基因研究识别人类疾病的致病基因可以为这些疾病的检测、治疗和预防提供重要的见解。在寻找人类复杂疾病的潜在基因时，基于单倍型的关联分析已被认为是一种具有高分辨率的工具，更重要的是，它在识别基因的适度病因学影响方面具有潜在的巨大能力。然而，在实践中，其效果并不像理论上预期的那样成功；一个主要原因是，这种分析需要大量的参数，以捕获丰富的单倍型品种。虽然高度的自由度可能妨碍确定对复杂疾病的适度遗传影响的能力，但需要纳入其他风险因素的协变量，这进一步恶化了绘制复杂疾病基因的自由度问题。为了解决这一问题，本文构建了一个高效、强大的基于模型的单倍型关联分析框架。该方法开发的核心重点是在基于模型的框架中有效利用单倍型信息，并在不同的研究阶段考虑不同的降低单倍型复杂性的策略以优化效率。对于筛选阶段的分析，PI构建了基于回归平台中成对比较的单倍型相似性的方法，以检测可能携带疾病基因的染色体区域。对于精细化阶段的分析，PI开发了基于进化的单倍型分组方法，以识别特定疾病相关的单倍型，并将新的降维技术集成到现有的回归方法中。本研究不仅为复杂基因定位提供了新的关联工具，而且为病例对照研究提供了一种常规方法，为今后的研究奠定了方法学基础。有了更好的统计工具，科学家可以提高他们对复杂疾病的理解，并设计更好的诊断和治疗策略，以改善人类健康。

AbstractPrincipal Investigators: Tzeng, Jung-Ying Proposal Number: DMS-0504726 Institution: North Carolina State University Proposal Title: Haplotype-based Association Modeling for Whole-Genome Scan and Candidate Gene Studies Identifying genes responsible for human diseases can illuminate significant insight to the detection, treatment and prevention of these diseases. In the search for genes underlying human complex diseases, haplotype-based association analysis has been recognized as a tool with high resolution and, more importantly, potentially great power for identifying modest etiological effects of genes. However, in practice, its efficacy has not been as successful as expected in theory; one primary cause is that such analysis requires a large number of parameters in order to capture the abundant haplotype varieties. While high degrees of freedom can hinder the power of identifying modest genetic effects on complex diseases, the need to incorporate covariates of other risk factors further worsens the degrees-of-freedom problem in mapping genes for complex diseases. To tackle this issue, the proposed work constructs an efficient and powerful model-based framework for association analysis at haplotype level. The central focus of the methods development is on the efficient use of haplotype information in a model-based framework, and different strategies of reducing haplotype complexity are considered at different research stages to optimize efficiency. For screening-stage analyses, the PI constructs approaches based on haplotype similarity of pair-wise comparison in a regression platform to detect chromosomal regions that are likely to harbor disease genes. For refinement-stage analyses, the PI develops evolutionary-based methods of haplotype grouping to identify specific disease-associated haplotypes, and integrate new dimension-reduction techniques into existing regression methods.The proposed work aims not only to provide novel association tools in complex gene mapping, but also develop a routine method for case-control studies and offer a methodological foundation for future advancement. With the availability of a better statistical tool, scientists can advance their understanding of complex diseases, and design better diagnostic and therapeutic strategies that improve human health.