基因表达调控是彼此联系相互制约的，构成了复杂的调控网络，是后基因组时代研究的重要课题之一。本项目将利用新一代高通量测序技术，从跨组学的角度来研究TFs、microRNAs、ceRNAs以及对应靶标基因间的多因素调控关系，提出一个基于因子图的基因调控网络建模和概率推断体系。研究内容包括建模前数据的收集和预处理、基因调控网络的初始构建、基于因子图的网络结构概率推断和算法实现、基于网络的拓扑结构统计分析和基因功能富集分析，并以PMA诱导的白血病K562细胞系为应用实例，获得不同分化条件下的K562跨组学数据（mRNA-Seq, miRNA-Seq和ChIP-Seq），运用上述模型体系来研究该细胞的基因调控机制。本项目的开展除了给具体生物过程、特定疾病或某种特定组织提供了一套研究多因素调控网络的模型体系和实用工具外，也为跨组学水平上认识造血细胞分化的基因调控网络提供了新的线索。

Gene regulatory networks have an important role in every biological process of life. New generation DNA sequencing provides several powerful platforms for analyzing the gene regulatory networks inside cells on the whole genomic level, such as mRNA-Seq, miRNA-Seq and ChIP-Seq etc. However, to understand the whole stories of gene regulation inside a cell, we have to integrate the different omics data together. By considering multi-factors of TFs, miRNAs, ceRNAs and target genes from cross-omics data, we mainly present a probabilistic framework, based on the factor graph, to model the gene regulatory network and infer its structure using RNA-Seq and miRNA-Seq profiles. The major tasks in this proposal include: (1)Data collection and pretreatment, find candidate interactions among TFs, miRNAs, ceRNAs and target genes, construct an initial regulatory network; (2) Define the probabilistic graph model and develop its learning procedure based on mRNA-Seq and miRNA-Seq profiles; (3) Use the learned model to infer the gene regulatory network of PMA-induced cell K562 based on its sequencing data; (4) Topological characterization and gene function enrichment analysis of the inferred network to get biological insights about its underlying regulatory mechanism. It is desired that our efforts could draw a comprehensive view of multi-factor gene regulatory network by integrating cross-omics sequencing data. We anticipate that our introduced probabilistic framework for gene regulatory network construction could inspire more and more applications as more and more cross-omics data becomes available in the very near future.