Statistical methods in genetic studies

在人类遗传学中，最终的研究目标是通过基因研究来了解复杂系统背后的生物学机制，以便人类能够从这些研究活动中获得的知识中受益。我的研究项目的共同主题是开发统计方法来整合不同类型的基因组数据(例如，遗传标记数据、微阵列数据)，用于不同目的的遗传研究。我的研究主题之一是开发统计方法，用于在复杂疾病基因定位的遗传连锁研究中使用密集分布的标记。当密集间隔标记用于连锁分析时，紧密间隔标记之间往往有很强的关联性。对协会的不当处理可能会导致严重的偏颇结果。为了解决这一问题，笔者提出了一种将关联效应融入到现有的连锁分析方法中的统计方法。另一方面，我提出了基于图形模型来刻画标记之间的关联模式。连锁分析领域的另一项研究重点是应用高斯过程的水平交叉理论来模拟ASP分析中的漂移峰过程，以检测包含疾病基因的染色体。尽管生物技术可以低成本地对单个核苷酸多态(SNP)进行基因分型，但当研究数千个受试者时，对大量SNP进行单个基因分型是昂贵的。将个体DNA汇集在一起进行基因分型一直被提倡，因为它可以有效地减少成本、劳动力和时间。汇集方法的挑战集中在从汇集在一起进行基因分型的一组个体中估计等位基因频率的精确度。由于同时汇集两个个体可以将基因分型成本降低两倍，我们建议开发一种新的基因分型算法，用于确定当两个个体被汇集进行基因分型时，两个个体对所拥有的等位基因数量。

In human genetics, the ultimate research goal is to understand the biological mechanism underlying the complex system through genetic studies, so that, human beings can benefit from the knowledge that derives from these research activities. The common theme of my research projects is to develop statistical methods for integrating diverse types of genomic data (e.g., genetic marker data, micro-array data) for genetic studies in different purposes.One of my research topics focuses on developing statistical methods for using densely spaced markers in genetic linkage study for complex disease gene mapping. When densely spaced markers are used for linkage analysis, there are often strong associations among tightly spaced markers. Improper handling of the association may lead to significantly biased results. I propose a statistical method that incorporates the association effects into the current existing linkage analysis method to solve this problem. On the other hand, I propose to characterize the association patterns among markers based on graphical models. Another research topic in the area of linkage analysis focuses on applying the theory of level crossings of a Gaussian process to model the excursion peak process in ASP analysis for detecting the chromosome that contains disease gene.Although biological techniques allow genotyping a single nucleotide polymorphism (SNP) at low cost, individual genotyping of a large number of SNPs is expensive when thousands of subjects are studied. Pooling individuals' DNA together for genotyping has been advocated as it can efficiently reduce cost, labor and time. The challenge of the pooling approach focuses on the precision of allele frequency estimate from a group of individuals pooled together for genotyping. As the pooling of two individuals at one time may reduce the genotype cost two fold, we propose to develop a new genotyping algorithm for determining the number of alleles possessed by the pairs of individuals when two individuals are pooled for genotyping.