关于基因与基因（环境）交互作用推断问题的若干统计方法研究

With the advances in modern biological technology, there are more than 1000 single nucleotide polymorphisms (SNPs) were identified with the association of complex diseases, however, it was proved that the identified SNPs only explained an small part of the phenotypic variance. One reason is probably that the current research methods always treat all SNPs one by one in the genome-wide association studies, while neglecting the gene-gene and gene-environment interactions. As a result, much information is lost. Some research results clearly pointed out the existence of these interactions. This project will check and summarize the existing statistical methods for testing these interactions at first, and then propose reasonable statistical methods than the existed methods. This project will also propose more efficient or robust statistical methods for testing a lot of interactions. The results obtained by this project will allow researchers to test gene-gene and gene-environment interactions more efficiently. At last, we will provide the computation package of the proposed procedures for users in the related fields. All statistical methods proposed by this project will be applied to the analysis of the actual datas.

在全基因组关联研究中，已经有上万个位点单核苷酸多态性与复杂疾病关联被识别。但是，这些被发现的位点只能解释很小一部分表型变异，原因之一是目前全基因组关联分析大多是对每个位点单独检验，而忽视基因与基因（环境）的交互作用，导致很多信息丢失。大量研究成果明确指出这种交互作用确实存在。现有关于基因交互作用的统计推断方法有很多局限性，如当环境变量取值较复杂时，现有统计推断方法精度不高，检验功效偏低；当多个交互作用同时检验时现有统计检验方法功效也不稳健。本项目拟在分析、总结现有关于基因交互作用统计推断方法的基础上，提出更加精确、更加符合实际的统计推断方法，针对多个基因交互作用同时检验问题拟提出功效更高或更稳健的统计方法。本项目拟完成适合新统计推断方法的计算程序包，供相关领域实际研究者使用。本项目所有的方法拟都应用于实际数据的分析。

随着现代生物学技术的发展，已经有千余位点单碱基多态性与复杂疾病关联被成功识别。但是，这些被发现的位点只能解释很小一部分表型变异，原因之一是目前全基因组关联分析大多是对每个位点单独检验，而忽视基因-基因与基因-环境的交互作用，导致很多信息丢失。大量研究成果明确指出这种交互作用确实存在。本项目在比较、分析、总结现有关于基因交互作用统计推断方法的基础上，提出更加精确、更加符合实际的统计推断方法，针对多个基因交互作用同时检验问题拟提出功效更稳健的统计方法。 通过模拟研究，证实了这些方法的稳健性，本项目研究所取得的方法都应用于某些实际数据的分析中。