偏最小二乘路径模型在皮肤老化遗传分析中的构建与应用

The genetic basis of complex traits is currently a hot topic in biological studies, but remains a big challenge. In previous studies, the applicant had constructed two methods for detecting genes and gene-gene interactions associated with complex traits based on multiple loci, which improved the analytic power. However, complex traits are not only results of multiple-gene interactions, but the correlation among phenotypes are also complicated. Taking skin aging as an example, it constitutes of pigment spots sizes and numbers, wrinkles and other phenotypes in different parts of the body, among which the correlation is complicated. We previously applied a partial least square path model into the skin aging study, and extracted a few composite phenotypes. Each of the composite phenotype encompasses multiple correlated phenotypes and demonstrates a specific physiological status (e.g. pigmentation on the face). But it still lacks efficient methods dissecting the genetic architecture of complex traits from a ‘multiple-loci multiple phenotypes’ perspective. In this project, the applicant will construct a new model for delineating the relation among multiple genes and multiple phenotypes. Specifically in the model, composite phenotypes will be extracted from initial phenotypes, and then the remaining part of initial phenotypes will be calculated; epistasis of biological meaning will be constructed in the model; and the relations among genes, epistasis, composite phenotypes and the remaining part of initial phenotypes will be estimated and tested in the model. The new model will be applied in the skin aging study to cast a new light on relations among multiple genes and multiple phenotypes. This project will provide a new method for association studies of complex traits.

复杂性状的遗传机制一直是生物学领域的研究重点和难点。申请人在前期工作中构建了基于多位点及交互作用的统计方法，提高了复杂性状关联分析的效力。但是，复杂性状不仅是多基因互作的结果，而且性状之间关系复杂。如皮肤老化，表现为身体各部位的色斑大小、色斑数量、皱纹等多个表型。每个表型都是由多基因决定的，且表型之间关系复杂。申请人前期将偏最小二乘路径模型应用到皮肤老化多表型分析，提取若干复合表型因子，每个因子对应明确的生理特征（如面部色斑），效果较好。但是，目前尚缺乏从“多基因-多表型”的角度系统研究复杂性状遗传机制的方法。本项目拟构建基于偏最小二乘路径模型的“多基因-多表型”交互作用网络分析方法，提取复合表型因子，计算复合表型因子提取后的表型剩余部分，构建生物学意义的交互作用，系统地研究多基因多表型之间的关系。本项目为复杂性状遗传分析提供新方法，并系统解析皮肤老化多基因多表型之间的关系。

本项目构建了基于偏最小二乘路径模型的“多位点-多表型”交互作用网络分析方法，不仅可以从多表型中提取若干复合表型因子，而且包含了具有生物学意义的交互作用。基于偏最小二乘路径模型的复合表型提取方法，在全基因组关联分析研究中已经被证实与单表型全基因组关联分析相比，其检验效能更高。本项目构建的基于偏最小二乘路径模型的“多位点-多表型”交互作用网络分析方法，在理论上具有更好的表现以及明确的生物学意义。在皮肤老化研究中，基于偏最小二乘路径模型提取复合表型之后进行全基因组关联分析找到了单表型全基因组关联分析没有发现的显著信号。这充分体现了基于偏最小二乘路径模型在全基因组关联分析中的优势。进一步，我们利用皮肤老化数据评估所构建的基于偏最小二乘路径模型的“多位点-多表型”交互作用分析方法在实际应用的表现。结果发现本项目提出的“多基因-多表型”交互作用分析方法不仅可以有效检测出传统的交互作用效应，即基于加性效应之间的交互作用，还可以检测出除此之外的交互作用，如第一个位点与第二个位点之间的加性效应与显性效应，显性效应与显性效应，显性效应与加性效应的交互作用均可以有效地检测。我们构造的“多基因-多表型”交互作用分析方法具有更全面的生物学含义，对于探索交互作用对表型的影响更加有效。