MSFBNP: Drug Repositioning Based on Multi-similarity Fusion and Bipartite Network Projection

Drug repositioning helps reduce the time and cost of drug research and development. In this study, we propose a novel method, called MSFBNP, to find potential indications for existing drugs. First, based on various biological data, multiple drug-drug similarity and disease-disease similarity are measured. Then, these similarities are further fused to improve drug-drug similarity and disease-disease similarity. Next, a cluster method is conducted to adjust the improved similarity for drug and disease, respectively. On the basis, two new adjacent matrices are constructed. Finally, two-step diffusion approach is adopted to predict new potential drug-disease association. The experimental results show that MSFBNP outperforms the other state-of-the-art methods. In addition, case studies for selected drugs further confirm the favorable prediction performance.