Drug-target interaction prediction using Edge2vec Algorithm on the heterogeneous network via SVM
Fatemeh Fattahi, Mohammad S. Refahi, Behrouz Minaei Bidgoli · 2019
Drug-target protein interaction (DTI) prediction is considered as a fundamental step in the process of drug discovery. The experimental methods in this domain, despite having accurate and beneficial results, are time-consuming. In the last few years, not only have these computational methods been overcome to these problems, but the results of them have been extremely efficient. One type of these techniques is based on network prediction on biomedical networks. Today, different network-based approaches have been suggested for identifying relations between drugs and targets which work on homogeneous networks well. But these approaches do not consider edge semantics in the network. In this paper, we employ the edge2vec algorithm for node embedding on the heterogeneous network and then by the use of the support vector machine method to predict drug-target interactions. Our method has obtained a significant accuracy than the existing methods.