CHADTI: A Sequence-Based Model with Attention Mechanism for Drug-Target Interaction Prediction

Abstract

In the new drug development process, drug discovery and repositioning can significantly reduce development costs and shorten development time. Drug-target interactions(DTIs) prediction is a crucial step for in-silico drug discovery and drug repurposing. However, manual identification of DTIs is extremely costly and time-consuming due to the large-scale chemical space. Using virtual screening to identify DTIs can significantly shorten development time and reduce costs. Recently, various attention models have been used to extract the features of drugs and proteins. However, these methods ignore the complex interactions between drugs and proteins. In this paper, we propose a sequence-based model, CHADTI, for DTIs prediction. We use stacked 1D-CNN layers to learn the features of drugs and a pre-trained model(ESM) to embed proteins. We design a novel attention mechanism to extract the complex interactions between drugs and proteins. Finally, We evaluate CHADTI on two datasets, and the results show that our model achieves state-of-the-art performance.