Prediction of drug-protein interaction network based on preferential attachment

Abstract:
Drug-protein interaction (DPI) network prediction plays an indispensable role in the discovery of new drug functions. In this paper, we used only the interactions between drugs and proteins to build networks that no require additional information about drug and protein. This network shows the specificity of drug and protein binding. We used the preferential attachment (PA) method to explore whether networks with multidrug proteins have similar social network characteristics. The results show that the DPI network has a Matthew effect similar to social networks. If the protein has a more ligand interaction, the protein is more likely to be bound by other drugs. In the experimental results, we found that the AZM (Acetazolamide) could bind to trypsin, and ABN (Benzylamine) could bind to HIV-1 protease binding, which was successfully predicted by the PA method.