GSL-DTI: Graph structure learning network for Drug-Target interaction prediction

•We propose an automated end-to-end graph structure learning model, GSL-DTI, for Drug-Target Interaction (DTI) prediction. In contrast to previous studies relying on manual rules, our approach incorporates an automatic graph structure learning method, which utilizes a filter gate on the affinity scores of DPPs and relies on the classification loss of downstream tasks to guide the learning of the underlying DPP network structure.•We conduct experiments on three public datasets and compare our method against competitive baselines. The experimental results demonstrate a significant outperformance of our model over state-of-the-art methods.•Furthermore, the introduction of graph structure learning offers a fresh perspective for DTI prediction research. To the best of our knowledge, GSL-DTI represents the first attempt to apply automatic graph structure learning to DTI tasks. It reduces the reliance on expert knowledge and yields improved node representations for downstream tasks.