Research On Prediction Of LncRNA-disease Association Based On Multi-similarity Network Fusion

More and more studies and experiments have shown that long non-coding RNA(lnc RNA)plays a key role in the occurrence and development of various human biological processes,and is linked to a variety of complex diseases.Analyzing the potential relationship between lnc RNA and disease and exploring the "role" played by lnc RNA in the development of disease is of great significance for understanding the pathogenic mechanism of disease at the molecular level and for disease prevention,diagnosis,and clinical treatment.In recent years,lnc RNA-disease prediction models based on machine learning algorithms have been proposed.Such models usually use known lnc RNA biological characteristics and disease information to train supervised or semi-supervised classifiers to predict potential association relationships.However,some of these methods require high-quality negative sample data.In reality,negative sample data is difficult to obtained.In addition,prediction models based on biological networks are often used in the prediction of lnc RNA-disease associations.Among them,the data fusion-based solutions often achieve superior performance results.Now,lnc RNA-disease association prediction models based on data fusion in bioinformatics generally achieve superior performance.But some messages would be lost and the weight ratios of different types of data sources are ignored in the process of integrating multi-source data of lnc RNA or diseases into homogeneous data.We are inspired to propose a novel method called RWSF-BLP.Firstly,lnc RNA and disease networks are integrated through a multi-similarity fusion technique for thefusion of additional available information.Then,the final correlation scoreis calculated using the bidirectional label propagation method.The experimental results showed that RWSF-BLP achieved 0.9086 and 0.9115＿0.0044 in LOOCV and 5-fold-CV.Moreover,case studies on three common diseases(leukemia,lung cancer and colorectal cancer)further demonstrated the predictive power of RWSF-BLP.Thus,we can conclude that RWSF-BLP can accurately infer potential lnc RNA-disease associations and can serve as a powerful tool for lnc RNA-disease associations prediction.