Effective prediction of drug-target interaction on HIV using deep graph neural networks

Individuals infected with HIV are controlled by drugs known as antiretroviral therapy by suppressing the amount of HIV in the body. Therefore, studies to predict both HIV-drug resistance and virus-drug interaction are of great importance for the sustainable effectiveness of antiretroviral drugs. A solution to this problem is provided by investigating the connection between the recently emerging geometric deep learning method and the evolu-tionary principles governing drug resistance to the HIV. In this study, geometric deep learning (GDL) approach is proposed to predict drug resistance to HIV, and virus-drug interaction. The drug data set in the SMILES repre-sentation was first converted to molecular representation and then to a graph representation that the GDL model could understand. Message Passing Neural Network (MPNN) transfers the node feature vectors to a different space for the training process to take place. Next, we develop a geometric neural network architecture where the graph embedding values are passed through the fully connected layer and the prediction is performed. The obtained results show that the proposed GDL method outperforms existing methods in predicting drug resistance in HIV with 93.3% accuracy.