Abstract:MicroRNAs have been reported to have close relationship with diseases due to their deregulation of the expression of target mRNAs. Detecting disease-related microRNAs is helpful for disease therapies. With the development of high throughput experimental techniques, a large number of microRNAs have been sequenced. However, it is still a big challenge to identify which microRNAs are related to diseases. Recently, researchers are interesting in combining multiple-biological information to identify the associations between microRNAs and diseases. In this work, we have proposed a novel method to predict the microRNA-disease associations based on four biological properties. They are microRNA, disease, gene and environment factor. Compared with previous methods, our method makes predictions not only by using the prior knowledge of associations among microRNAs, disease, environment factors and genes, but also by using the internal relationship among these biological properties. We constructed four biological networks based on the similarity of microRNAs, diseases, environment factors and genes, respectively. Then random walking was implemented on the four networks unequally. In the walking course, the associations can be inferred from the neighbors in the same networks. Meanwhile the association information can be transferred from one network to another. The results of experiment showed that our method achieved better prediction performance than other existing state-of-the-art methods.

A Novel Method of Predicting Microrna-Disease Associations Based on Microrna, Disease, Gene and Environment Factor Networks.