Research Of Coregulated Genes Mining Technology Based On Data Fusion

21st century is dominated by biotechnology and is also the century of life scienceand information technology. Mining co-regulated genes has become the core area ofresearch about gene function association. Novel DNA microarray technology has beenable to simultaneously monitor the expression levels of thousands of genes, As one of thebioinformatics data sources, gene expression data is already in an increasingly importantposition. Analyzing how regulatory factors regulate the expression of certain genes canhelp us use microarray data to construct regulatory networks and dig co-regulated genes,thus constructing gene regulation module is one of the key challenges. Co-regulatedgenes are defined as a group of genes that are regulated by the same transcription factoror set of factors in an organism. Gene expression level can be regulated at thetranscriptional level, translational level or many other stages. Construction of generegulatory networks and dig of co-regulated genes allows us to understand physiologicalactivities and functions of the organism form the molecular level and to understand howorganisms differ according to the expression of the gene produce changes.The most common excavation method of regulated genes is calculatingco-expression genes from microarray data and achieving transcriptional control module.However, in addition to similar expression level co-regulated genes can show, but mostof them also have similar functions that work together to complete the living process. Bycombining the expression data and the ontology, we can analyze and discover therelationship between genes more accurately and efficiently.In this paper, in order to mine co-regulated genes, we analyze the expression profilesimilarity, ontology similarity and integrated network set respectively, so the studyconsists of three aspects:(1) By studing gene expression profiles in co-regulated genes,we propose a newsimilarity measure based on subspace search.It firstly searches for the spatial correlationof the gene, and then calculate the similarity of the gene expression levels in thesubspace.Thus can better reflect the spatial information of gene expression.(2) The gene ontology enables us to use richer information to predict gene’scorrelation function.Scholars compared gene’s function using its structure andannotateion which is often referred to as the semantic similarity between genes. Throughanalyzing the structure of gene ontology, we proposed a semantic similarity measure forgene function based on semantic differentiation.This method introduces the idea ofsemantic differentiation factor and shortest path and redefined the information containedin each term so that each term throughout the GO graph has a fixed amount ofinformation. The semantic differentiation factor is decided by the number of descendants the node has. This approach combines the path between nodes and information, fullyconsiders the location information on the terms in the topology structure, can better reactthe relationship between organism.(3) Combining expression data with ontology data and then use integratedinformation to construct regulatory networks.we apply frequent dense subgraph miningtechnology in the graphset to dig co-regulated genes, The result shows that this methodhas good potential applications, gene regulatory networks with data fusion can beconstructed to provide valuable information.