A Novel Approach for Predicting LncRNA-Disease Associations by Structural Perturbation Method

Abstract

Numerous experiments have demonstrated that long non-coding RNA (lncRNA) plays an important role in various systems of the human body. The prediction of lncRNA-disease associations is conducive to the diagnosis and prevention of complex diseases. However, the number of known disease-associated lncRNAs is very small. Therefore, predicting the associations between lncRNAs and diseases by computational models has become an urgent need. In this paper, we propose a model called SPMLDA (Structure Perturbation Method LncRNA-Disease Association), which establishes a bi-layer network by integrating disease similarity, lncRNA similarity and known lncRNA-disease associations. Then, we completed lncRNA-disease association matrix based on the structure perturbation model. SPMLDA obtained AUCs of 0.8823 ± 0.0034 based on LncRNADisease dataset and 0.8721 ± 0.0021 based on Lnc2Cancer dataset in the leave-one-out cross validation, which is higher than the state-of-the-art prediction methods. Finally, case studies of two complex human diseases further confirmed the superior performance of our model. SPMLDA could be an important resource with potential values for biomedical researches.