An Efficient Computational Method to Predict Drug-Target Interactions Utilizing Structural Perturbation Method.

Accurately and quickly identifying potential drug candidates for therapeutic targets (i.e., drug-target interactions, DTIs) is a basic step in the early drug discovery process. However, the experimental determination of drug-target interactions is expensive and time-consuming. Therefore, a continuous demand that the effective calculation algorithm is developed with a more accurate prediction of drug-target interactions. Some algorithms have been designed to infer new interactions but ignored the link generation mechanism in the DTI network, and the missing information can be completed by generalizing the observed DTI network structure according to some consistency rules. We propose a calculation model named SPDTI, which is based on the structural perturbation method and uses the explicit and implicit relations of the drug-target adjacency to predict novel DTIs. In our framework, we first construct the implicit relationship between nodes of the same type through the display relationship of the DTI adjacency matrix. Then a bilayer network is designed to integrate these two relationships, and finally, the structure perturbation method is used to predict the potential interactions between drugs and targets. To evaluate the performance of SPDTI, we carried out a lot of experiments on four benchmark datasets. The experimental results show that SPM can be used as an effective computing method to improve the identification accuracy of drug-target interactions.