Bioinformatics Database Construction, Analysis And Prediction Of LncRNA-gene Regulatory Relationships

As a new part of biomacromolecules, long non-coding RNAs (lncRNAs) were critical in various biological processes. With the development of lncRNA research, plenties of data related to lncRNAs emerged. In order to make better use of these information, tots of bioinformatics databases have been built. These databse contained structure information, expression information and interaction information of lncRNAs, and played an important part in the study of lncRNAs. Moreover, the data collected by these databases was useful for bioinformatics study of lncRNAs with machine leaning method.With the continuous development of lncRNAs, people found that lncRNAs played a crical part in the regulation of cell function. There was no database whick was foucsed on lncRNA-gene regulatory relationships, and it was inconvenient to obtain lncRNA regulatory information. To solve this problem, we collected lncRNA-gene regulatory relationships which were vertified by experimental evidences. Specifically, the structured data was normalized by analyzing the characteristics of lncRNAs related regulatory relationships. And then, regulatory relationships between lncRNAs and genes were collected by screening and analyzing full text related to lncRNAs mannully. Our work systematically organized the experimental research of lncRNA functional studies and provided a priori data for prediction of bcRNA-gene regulatory relationships with bioinformatics method. To store these lncRNA-gene regulatory relationships, a database called LncReg was built (http://bioinformatics.ustc.edu.cn/lncreg/). LncReg was a special designed database which contained regulatory relationships, regulatory mechanisms and other comprehensive and detailed information. Comparied with other databases, LncReg was more suitable to store lncRNA- gene regulatory relationships in databse structure and information collected in LncReg was more comprehensive and detailed. Based on regulatory relationships dataset collected in LncReg, lncRNAs regulatory networks were rebuilt and a series of analysis were carried out in this networks, such as topology analysis, pathway analysis based on KEGG and gene functional analysis based on Gene Ontology. Topology analysis showed that lncRNA-gene regulatory networks was in line with the typical characteristics of biological networks. Functional analysis showed that the regulatory relationships betweenlncRNA and gene were concentrated in cell cycle, tissue development and disease development.Finally, a prediction method called LRGP (LncRNAs Regulated Gene Prediction) was introduced to predict unknown lncRNA-gene regulatory relationships. Based on lncRNA-gene regulatory relationships dataset and combined with network-based inference, this method can predict potential lncRNA-gene regulatory relationship through resource allocation, and compared with previous method LRGP has a better perferacne on prediction accuracy.