Drug-Target Interaction Prediction Based on Gaussian Interaction Profile and Information Entropy

Lina Liu, Shuang Yao, Zhaoyun Ding, Maozu Guo, Donghua Yu, Keli Hu

Abstract: Identifying drug-target interaction (DTI) is an important component of drug discovery and development. However, identifying DTI is a complex process that is time-consuming, costly, long, and often inefficient, with a low success rate, especially with wet-experimental methods. In contrast, numerous computational methods show great vitality and advantages. Among them, the precisely calculation of the drug-drug, target-target similarities are their basic requirements for accurate prediction of the DTI. In this paper, the improved Gaussian interaction profile similarity and the similarity fusion coefficient based on information entropy are proposed, which are fused with other similarities to enhance the performance of the DTI prediction methods. Experimental results on NR, GPCR, IC, Enzyme, all 4 benchmark datasets show that the improved similarity enhances the prediction performance of all six comparison methods.