Drug-Target Interaction Prediction with Deep Learning and Recommender Systems

Drug designing is a complex, costly, and time-consuming process with high failure chance. Therefore, drugrepurposing is gaining importance. To this aim, one method is to identify new interactions between chemicalcompounds and protein targets called predicting Drug-Target Interactions (DTIs). DTIs are usuallyrepresented by bi-partite networks where predicting drug-target interaction can be formulated as the linkprediction. Recently, Graph Neural Networks (GNNs) have achieved tremendous success on machinelearning tasks defined over graph-structured data such as node classification and link prediction. They canextract informative features of the input graph and improve the performance of down-stream tasks.We propose a GNN-based framework to predict drug-target interactions. To this aim, we represent the drugtargetinteractions as a bi-partite network and we construct a protein-similarity network between proteinsbased on their structural similarities. The proposed framework learns informative representations for bothdrugs and target proteins in an end-to-end fashion. The learned representations are used to predictinteractions. We prepared two data sets. The first one extracted from DrugBank containing relevantinformation about drugs, target proteins, and their interactions. As the second data set, we used the Yamanishibenchmark dataset containing interactions between drugs and different groups of enzymes, ion channels,GPCRs, and nuclear receptors.The results indicate that the proposed framework exhibits acceptable performance and can get better resultscompared to some proposed methods in the literature. On the first data set, our method achieved ۹۲% and۸۵% of accuracy over training and test sets. For the second data set, accuracies are ۸۹%, ۸۶%, ۸۲%, and۸۰%, respectively, on four classes of targets.The proposed GNN-based framework starts with random representations for drug and proteins and learnshighly informative embeddings to predict the possible interactions. Results, indicate the high abilities of themethod in predicting DTIs.

کلیدواژه ها:

Graph Neural Network ، Drug-Target Interaction ، Drug Repurposing ، DTI Classification