Research On Functional Modules Structure Analysis And Module Recognizing Algorithms For Transcriptional Regulatory Networks

In order to better insight into the correlation between the topological structure andfunction of transcription regulation networks, in this paper, the thought of complexsystem and network science has been be used to analysis and research functionalmodules of transcriptional regulatory networks in two ways, which are from networkstructure to gene function and gene function to network structure.Transcriptional regulation is an important part of gene expression processes, whichis playing a crucial role in the genetic and evolutionary of biology. With the rapiddevelopment of network technology, the mechanism and information of transcriptionalregulation have been trying to reveal through analysis of the topology characters oftranscriptional regulatory network composed of gene-gene interactions, which based onthe complex network theory. Modularity is one of the basic properties of complexnetworks. Integrating modular structure with biological functions is an efficient methodto analysis of biological networks. There are two ways of the connections: one is toobtain network modules based on the known functional annotation and then analysisthe structure of them; the other is to identify functional modules based on topology ofnetworks. However, there is not unified standard for the division of functional modulesof transcriptional regulatory networks. Various databases have their own standard sothat the modules of different transcriptional regulatory networks can not compare. Inthe aspect of module identification, requires specific methods to identify high-degreeoverlapped functional modules in transcriptional regulatory networks to meet the goalof predict gene functions by means of network structure.In view of these problems and needs, our works presented a directory of functionalannotation which could be applied to prokaryotes and eukaryotes, and then extractedmetabolic gene transcriptional regulatory subnets and10functional module subnetsfrom E. coli and Yeast transcriptional regulatory networks based on this directory. Thetopology analysis of them showed several significant conclusions. In addition,according the topology characters of transcriptional regulatory networks, we proposed two algorithms to identify functional modules in the transcriptional regulatory networksof prokaryotes and an algorithm in eukaryotes. The main contains and innovations inthis paper includes five points:⑴We presented a hierarchical directory of functional annotation, which havewider use range than other past directories. Then we extracted more complete metabolicgene transcriptional regulatory subnets and10functional module subnets from E. coliand Yeast transcriptional regulatory networks based on this directory. The topologyanalysis of them showed that topological properties of metabolism gene transcriptionalregulatory subnets are similar with the transcriptional regulatory networks, but differentfrom the functional module subnets.⑵In order to detect functional modules in sparse transcriptional regulatorynetworks of prokaryote, we presented a new algorithm based on edges similarityOMDL. The algorithm reinvent modules as groups of links rather than nodes. We findrelevant link modules in the transcriptional regulatory network of E. coli. The resultsinfer that OMDL algorithm could detect functional modules in transcriptionalregulatory networks of prokaryote effectively.⑶To improve search rate and accuracy, we presented a new algorithm on thebasis of link network FOMDL. The algorithm converted the directed network to theundirected weights network. Links in the original network have been turned into nodesin link network. Then detecting communities in the link network could obtain functionalmodules from original network simultaneously. We find relevant modules in thetranscriptional regulatory network of E. coli. The results infer that FOMDL is moreaccurate and fast than OMDL.⑷In order to detect functional modules in dense transcriptional regulatorynetworks of eukaryotes, we presented a new algorithm on the basis of the distributioncharacteristics of triples OMDT. As the smallest subnet consisting of three nodes, thetriples distribution characteristics has its own regularity. Based on these characteristicswe calculated the similarity of edges and detect functional modules. We find relevant functional modules in the transcriptional regulatory network of Yeast. The results inferthat OMDT algorithm could detect functional modules in transcriptional regulatorynetworks of prokaryote effectively and more accurate than previous methods inidentifying overlapping nodes.In conclusion, we focus on the research on two aspects: analyzing the structurefeatures of transcriptional regulatory networks and detect functional modules intranscriptional regulatory networks. These studies provides a new train of thought forthe people to analyze this kind of question.