Dual Sparse Collaborative Matrix Factorization Method Based on Gaussian Kernel Function for Predicting LncRNA-Disease Associations

Abstract

With the development of science and technology, there are increasing evidences that lncRNAs are associated with diseases, so it is particularly important to find some novel lncRNA-disease associations (LDAs). Finding some novel lncRNA-disease associations is benefit to us in the treatment and prevention of diseases, but this process requires a lot of energy and time, so there is an urgent need to discover some new methods. In this paper, a dual sparse collaborative matrix factorization method based on gaussian kernel function (DSCMF) is proposed to predict novel LDAs. DSCMF is based on the traditional collaborative matrix factorization method. The lncRNA network similarity matrix is combined with the lncRNA expression similarity matrix, and the disease network similarity matrix is combined with the disease expression similarity matrix. Therefore, the Gaussian interaction profile kernel is added. To increase the sparsity, the L2,1-norm is also added. Finally, the AUC value is obtained by ten-fold cross-validation method as the evaluation indicator of the performance of this method. The simulation experiment is used to predict some novel associations. The experimental results prove that this method will be of great help to future research.