Identification of Small Molecule-miRNA Associations with Graph Regularization Techniques in Heterogeneous Networks.

Abstract

MicroRNAs (miRNAs) are significant regulators of post-transcriptional levels and have been confirmed to be targeted by small molecule (SM) drugs. It is a novel insight to treat human diseases and accelerate drug discovery by targeting miRNA with small molecules. Computational approaches for discovering novel small molecule-miRNA associations by integrating more heterogeneous network information provide a new idea for the multiple node association prediction between small molecule-miRNA and small molecule-disease associations at a system level. In this study, we proposed a new computational model based on graph regularization techniques in heterogeneous networks, called identification of small molecule-miRNA associations with graph regularization techniques (SMMARTs), to discover potential small molecule-miRNA associations. The novelty of the model lies in the fact that the association score of a small molecule-miRNA pair is calculated by an iterative method in heterogeneous networks that incorporates small molecule-disease associations and miRNA-disease associations. The experimental results indicate that SMMART has better performance than several state-of-the-art methods in inferring small molecule-miRNA associations. Case studies further illustrate the effectiveness of SMMART for small molecule-miRNA association prediction.