TripletMultiDTI: Multimodal representation learning in drug-target interaction prediction with triplet loss function

In drug discovery, drug-target interaction (DTI) plays a crucial role. Identifying DTI in a wet-lab experiment is time-consuming, labor-intensive, and costly. Using reliable computational methods to predict DTI mitigates the enormous costs and time of drug discovery. Deep learning-based methods for predicting DTI have recently gained more attention. In DTI, drug-related and target-related data come in various modalities, which leads researchers to utilize multimodal approaches. It is shown that a discriminative feature representation of the drug-target pair plays the main role in multimodal DTI prediction. To achieve this goal, we propose a new multimodal approach that utilizes triplet loss jointly with task prediction loss. The proposed approach is called TripletMultiDTI. The proposed approach has two main contributions: 1) a new architecture that fuses the multimodal knowledge to predict interaction affinity labels and 2) a new loss function based on the triplet loss to learn more discriminative representation. Triplet loss encourages clustering of feature space such that similar drug-target pairs have the same feature space and dissimilar drug-target pairs have different feature space. As a result of our experiments, we were able to improve prediction performance. To this end, the proposed approach is evaluated on three well-known datasets and compared with state-of-the-art multimodal approaches. According to the obtained results, we can perform better than comparable approaches.