高通量组学技术的发展与成熟，实现了将影响疾病发生、发展与转归的全组学标记映射到沿DNA→RNA→蛋白质→代谢物→复杂表型这一连续统上的分子网络中，对多层面组学标记交叉整合，在人群中推断跨组学交互致病网络及其效应，已成为研究热点。本项目提出了跨组学交互网络回归模型的理论框架，以网络节点间的积矩项表征边向量，构建无向网络/通路固定效应模型；基于节点效应和边效应的协方差信息，构建无向网络/通路混合效应模型；进而，依据道义图（moral graph）原理，将有向网络转化成无向网络，并保留原有向网络蕴含的方向信息，构建有向网络/通路固定和混合效应模型。统计模拟评价效应估计的无偏性和有效性、统计量的稳定性、检验效能和优劣性，理论证明其渐进分布，实例分析结直肠癌跨组学数据验证其实用性，以期构建跨组学交互网络回归模型及其统计推断方法，丰富和发展跨组学网络生物学理论，为跨组学生物医学研究提供高效的分析方法。

With the advanced high-throughput -omic platforms, the globolomics that affecting the disease occurrence, development and prognosis has been mapped onto biomolecule network locating on the continuum DNA→RNA→Protein→Metabolite→Complex phenotype. It has become one hot topic to integrate multi-omics biomarkers to infer the effect of cross-omics interaction network. This project firstly proposed the theory framework for constructing cross-omics network regression model, developed the fixed effect model for undirected network by representing the edge utilizing the product moment of its connected nodes, and constructed the corresponding mixed effect model using the covariance information. Then, based on the moral graph theory, we have transformed the directed network into undirected ones holding the direction information simultaneously, and constructed the fixed and mixed effect model for directed network respectively. Various simulations are conducted to assess the unbiasedness and efficiency of the parameter estimator, the stability and power for the developed statistics, and the theoretical distribution has also been derived. Furthermore, one cross-omics real data set for colorectal cancer has been analyzed. It is expected that this project can provide the methods for regression model of cross-omics interaction network, riched the cross-omics network biology theory and to provide the efficient tools for cross-omics biomedicine research.