复杂疾病的全基因组SNP互作网络构建与分析

Complex diseases are normally caused by interaction effects of multiple factors. Genome-wide SNP interaction pattern detection is an important method to uncover underlying genetic mechanisms of complex diseases. Nevertheless, currently there are few studies dealing with the detection of SNP interaction patterns from the viewpoint of SNP network construction. Our project, based on simulation data and real genome-wide SNP data of specific complex diseases, namely, age-related macular degeneration and ovarian cancer, focuses on detecting interaction patterns using the network analysis methods from the constructed SNP interaction networks. Main research contents of the project consist of four parts. They are the design of association measures, the construction of SNP interaction networks, the detection of interaction patterns and the biological explanation of detection results. The research methods to be adopted mainly include the significant and stable association measures based on the co-information theory, the weighted hyper-graph and its corresponding equivalent graph based on the hyper-graph theory, network analysis methods (such as the degree of each vertex, the distance of a combination of several vertices, the network permutation test, and many others), gene ontology/pathway enrichment analysis. Among them, the significant and stable association measures, as well as the weighted hyper-graph and its corresponding equivalent graph, are innovations of methods of our project. Expected results will contribute to the research of big data mining, network construction and interaction pattern recognition, and also help to explain underlying genetic mechanisms of age-related macular degeneration and ovarian cancer. Hence, our project and its results have good social and economic benefits in the areas of prevention and early warning, early diagnosis, personal treatment, drug development, and so on.

复杂疾病往往是由多种因素共同作用导致的。全基因组SNP互作模式识别是解析复杂疾病遗传机理的重要方法，然而从SNP互作网络构建这一角度识别互作模式的研究当前还比较少。项目以仿真数据和特定复杂疾病（年龄相关性黄斑变性和卵巢癌）全基因组SNP数据为基础，致力于构建全基因组SNP互作网络并利用网络分析方法识别互作模式。主要研究内容有关联测度的设计、SNP互作网络的构建、互作模式的识别和结果的生物学解释。研究方法包括基于共信息理论的显著并稳定测度、基于超图理论的带权超图及其等价图、网络分析方法（包括顶点度、组合顶点间距和网络置换检验等）、GO和Pathway富集分析，其中显著并稳定测度和带权超图是项目的方法创新。项目成果将有助于大数据挖掘、网络构建分析和互作模式识别等问题的研究，也有利于解释年龄相关性黄斑变性和卵巢癌的遗传机理，在疾病早期诊断、个性化治疗及药物研制等方面有较好的社会效益和经济效益。

复杂疾病往往是由多种因素共同作用导致的。全基因组SNP互作模式识别是解析复杂疾病遗传机理的重要方法，然而从SNP互作网络构建与分析这一角度识别互作模式的研究当前还较少。项目以仿真数据和年龄相关性黄斑变性全基因组SNP数据为基础，致力于构建全基因组SNP互作网络并利用网络挖掘方法识别互作模式。主要研究内容有关联测度的设计、SNP互作网络的构建、互作模式的识别和结果的生物学解释等。项目主要以高水平论文和软件著作权的形式给出。重要结果包括基于带权超图及引入虚拟顶点的等价图实现了SNP互作网络的构建、基于高密度子图挖掘方法实现了高阶互作模式的识别、基于压缩感知理论实现了高维SNP数据的特征降维、基于启发式群体智能算法（蚁群算法和微粒群算法）实现了高维SNP数据的搜索、基于共信息理论实现了多个关联测度的设计、基于GO分析及KEGG分析实现了识别结果的生物学解释等。以上项目成果有助于大数据挖掘、网络构建分析和互作模式识别等问题的研究，也有利于解释年龄相关性黄斑变性的遗传机理，在疾病早期诊断、个性化治疗及药物研制等方面有较好的社会效益和经济效益。