Prediction And Analysis Of MiRNA And Gene Regulatory Networks

With the development of DNA sequencing technology and computer technology,the study of relationship between diseases and genes have been gradually realized.With the help of high-throughput sequencing technologies,researchers can obtain much more gene expression data from both patients and normal persons.By means of computer technologies,researchers can use computer's enormous computing power and algorithms to analyze the obtained data.Gene expression data reach up to more than twenty thousand dimensions,which brings a big challenge to current research.This also increases the difficulty to access the relationship between genes and mi RNAs.In recent years,the relationship between genes and mi RNAs are becoming a hot spot of research in bioinformatics.Researchers have found that there are many common aspects among multiple diseases S,such as some genes have similar expression in multiple diseases.Therefore it is important to identify key mi RNAs and association between genes and key mi RNAs which is shared by multiple diseases.Pan Cancer project then comes into being in order to explore the common markers exists in cancer.The work in this paper in described as follows.1.This paper proposed an algorithm termed DCMM(DNA methylation,Copy Number,mi RNA,m RNA),which utilizes multiple dataset to discover complex mi RNA regulation network.DCMM algorithm uses the linear regression model to explore the mi RNA regulation network based on the data of DNA methylation,Copy Number,mi RNA and m RNA.Since adopting of DNA methylation data and Copy Number data,DCMM estimates the relationship between mi RNA and m RNA more accurate compared with the existing methods,and the network constructed by our method is reliable.Localfdr algorithm and Cluster ONE algorithm are also used in DCMM algorithm to mine the underlying information.2.Twelve types of cancer data provided by TCGA is used in DCMM.DNA methylation dataset,Copy Number dataset,mi RNA dataset and m RNA dataset is contained in each cancer dataset.Based on these data,we identify mi RNA-m RNA matching pairs and construct mi RNA regulatory modules and mi RNA regulatory networks.3.In order to optimize the result,the mi RNA regulatory network is combined with protein-protein interaction network.Their intersection parts are selected to act as the regulation models.By analyze the models with existing database,we find that there are direct or indirect relationships between mi RNA-m RNA network substructures,which shows the effectiveness of DCMM.DCMM algorithm also can be used for analysis of other complex diseases.It provides reference for biological target therapy of polygenic diseases,and could clarify the pathogenesis and promote risk prediction.