A new method on lncRNA-disease-miRNA tripartite graph to predict lncRNA-disease associations

Finding the potential functions of lncRNAs is really vital for further study of human complex diseases. It requires a long time and other resources to uncover the potential lncRNA-disease associations by biological experiments, so developing computational methods to predict lncRNA-disease associations has become a hot topic in recent years. The prediction methods can basically rely on known lncRNA-disease associations or multitypes of data and molecular interaction networks. In this paper, we employ a method based on known lncRNA-disease associations, known disease-miRNA associations and validated lncRNA-miRNA interactions to construct a lncRNA-disease-miRNA tripartite graph and apply a modified resource allocation process to predict lncRNA-disease associations. In comparing with other related methods, our method achieves better performance with AUC and AUPR values of 0.984 and 0.828, respectively. Additionally, our method can predict latent lncRNA-disease associations for isolated lncRNAs or diseases.