Analyses Of Hypertension-Related Genes Based On Complex Network Theory

The study of complex systems clearly shows that the global behaviors of systems are determined by their structures rather than by properties of their individual parts. The complex network approach has become a powerful tool for studying complex systems, and the global properties of systems are usually studied by abstracting individual elements of systems into nodes and reducing interactions between elements to edges between nodes. Such an approach has been applied to understanding gene functions in biological systems.Essential hypertension is a disease caused by long-term interaction between genetic and environmental factors, and salt is one of the important environmental factors. The blood pressure response to salt loading or salt restriction is heteroge-neous among individuals, which is known as salt sensitivity. Salt sensitivity is the genetic susceptibility of individual blood pressure response to salt, and is an inter-mediate phenotype of essential hypertension. The clinical research and treatment of hypertension have improved dramatically. However, its molecular mechanisms and pathologies involved still need further study.In this thesis, based on both biological knowledge and network theory, a complex network approach is used to analyze the relationships among genes, path-ways and hypertension from different views by constructing several different net-work models (expression network, two-mode network and weighted networks) of hypertension-related genes. The approach is helpful to explore the complexity and diversity of the genetic factors and pathogenesis of hypertension, which pro-vides another perspective and a new idea for exploring molecular mechanisms and disease pathologies of salt-sensitive (SS) hypertension and other complex diseases.The main results of this thesis are listed as follows:1. Based on microarray data of the Dahl SS rat and two consomic rat strains, we reversely obtain the relationships between the genes by analyzing the corre-lations of their gene expression data, and construct the expression network of hypertension-related genes. The statistical and topological characteristics of the network show that it is sparse, small-world, scale-free and assortative. Combining with the analysis of three centrality indices (degree centrality, betweenness cen-trality and closeness centrality), we introduce a new integrated centrality index to comprehensively and quantitatively reflect the contribution of the three centrality indices. Then,16hub nodes of the gene expression network are determined, which typically correspond to key genes (Col4a1, Lcn2, Cdk4, etc.) playing important roles in hypertension and have been confirmed by biological/medical research.2. Through digging relevant information of genes and pathways, we construct a two-mode gene-pathway network that contains two different types of nodes. The statistical and topological characteristics of the network show that both gene nodes and pathway nodes have power-law degree distributions. A few genes and pathways with high degree are closely related to hypertension. In order to obtain the relationships between the same type nodes, we also construct a pathway-based gene network and a gene-based pathway network, respectively. The results show that these two networks have small-world properties and modular structures. The weak connections of the networks can be visualized by modular structures, which can help to filter out key hypertension-related genes or pathways. 3. Based on the above-mentioned two unweighted networks (gene network and pathway network), we further construct weighted gene networks and weighted pathway networks including edge-and node-weighted networks, respectively. It is known that weighted networks have more information comparing with unweighted networks:edge weights can describe the difference of interactions between nodes, and node weights can reflect the importance of nodes in networks. Therefore, by visualization of weighted networks, the important nodes and edges (connections) that correspond to key genes and pathways can be visually displayed, which have significant impacts on hypertension.