A Directed Weighted Network-Based Method for Drug Combinations Identification Using Drug-Target and Inter-target Regulation

Background Drug combination is currently a promising solution in treating complex diseases due to its reducing toxicity and enhancing therapeutic efficacy. However, the accurate identification of drug combination effects remains challenging. Results In this work, we propose a novel directed weighted network-based approach to identify drug combinations. Specifically, the network is constructed on both drug-target and inter-target interactions, together with their directed regulation. The biological processes of drug effects propagation and attenuation are modeled, aiming to capture direct and indirect drug actions on targets. By assigning weights to nodes of regulatory effects, relative distances between node sets within network can thus be computed. These distances are then analyzed to discriminate the combinatorial efficacy of various drug combinations. Empirical evaluations validate a remarkable working performance of the proposed method. Compared to existing approaches, our method is a better alternative on the task of drug combination prediction. Conclusion The proposed method reports a creative and practical scheme for identifying drug combination effects. With the analysis of drug-target and inter-target regulatory relation, our method is more competitive in distinguishing the combinatorial efficacy, which mitigates the deficiencies of classical drug combination prediction models.