Predicting of microbe-drug associations via a pre-completion-based label propagation algorithm
Haochen Zhao, Guihua Duan, Botu Yang +2 more
Identifying microbe-drug associations is important to systematically understand a drug’s mechanism of action in the therapeutic application. Since identifying microbe-drug associations is expensive and time-consuming via biological experiments, in this study, we propose a Pre-completion-based Label Propagation (PLP) method (called PLPMDA) to predict microbe-drug associations based on the multi-type similarities. To obtain richer information of drugs and microbes, we calculate drug chemical structure similarity, drug Anatomical Therapeutic Chemical (ATC) code similarity, microbe functional similarity, microbe sequence similarity and Gaussian Interaction Profile (GIP) kernel similarities of microbes and drugs, and then introduce a non-linear similarity fusion method. Comparing baseline methods, our advantage lies in performing an effective pre-completion step on the initial association…