AutoDTI++: deep unsupervised learning for DTI prediction by autoencoders

• DOI: 10.1186/s12859-021-04127-2
• Publication Date: 2021-04-20T09:03:36Z
• Authors: Seyedeh Zahra Sajadi; Mohammad Ali Zare...; Sajjad Gharaghani; Karim Abbasi
• Keywords: QH301-705.5; Research; Computer applications to medicine. Medical informatics; 0206 medical engineering; R858-859.7; Deep learning; Latent feature; Denoising autoencoder; 02 engineering and technology; Unsupervised learning; 3. Good health; Drug Development; Drug Discovery; Amino Acid Sequence; Biology (General); Drug-target interactions; Algorithms; Unsupervised Machine Learning

AbstractBackgroundDrug–target interaction (DTI) plays a vital role in drug discovery. Identifying drug–target interactions related to wet-lab experiments are costly, laborious, and time-consuming. Therefore, computational methods to predict drug–target interactions are an essential task in the drug discovery process. Meanwhile, computational methods can reduce search space by proposing potential drugs already validated on wet-lab experiments. Recently, deep learning-based methods in drug-target interaction prediction have gotten more attention. Traditionally, DTI prediction methods' performance heavily depends on additional information, such as protein sequence and molecular structure of the drug, as well as deep supervised learning.ResultsThis paper proposes a method based on deep unsupervised learning for drug-target interaction prediction called AutoDTI++. The proposed method includes three steps. The first step is to pre-process the interaction matrix. Since the interaction matrix is sparse, we solved the sparsity of the interaction matrix with drug fingerprints. Then, in the second step, the AutoDTI approach is introduced. In the third step, we post-preprocess the output of the AutoDTI model.ConclusionsExperimental results have shown that we were able to improve the prediction performance. To this end, the proposed method has been compared to other algorithms using the same reference datasets. The proposed method indicates that the experimental results of running five repetitions of tenfold cross-validation on golden standard datasets (Nuclear Receptors, GPCRs, Ion channels, and Enzymes) achieve good performance with high accuracy.