Drug-target interaction prediction using knowledge graph embedding.

Area of Science:

• Computational Biology
• Bioinformatics
• Drug Discovery

Background:

• Drug-target interaction (DTI) prediction is vital for drug development and repositioning.
• Existing knowledge graph methods for DTI prediction face challenges with data sparseness and incomplete information.
• Latent type information in knowledge graphs is often overlooked by current DTI prediction models.

Purpose of the Study:

• To propose TTModel, a novel knowledge graph embedding model for improved DTI prediction.
• To address limitations of existing methods by incorporating biomedical text and type information.
• To enhance the accuracy and robustness of computational DTI prediction.

Main Methods:

• Developed TTModel, a knowledge graph embedding model specifically for DTI prediction.
• Utilized biomedical text semantics to enrich the knowledge graph representation.
• Incorporated latent type information to improve the learning of node embeddings.
• Evaluated the model on two public datasets for DTI prediction.

Main Results:

• TTModel significantly outperforms state-of-the-art methods in DTI prediction.
• The model demonstrates improved performance by effectively learning from text semantics and type information.
• Experimental results validate the efficacy of TTModel on benchmark datasets.

Conclusions:

• TTModel offers a superior approach for predicting drug-target interactions.
• Integrating text and type information is crucial for enhancing knowledge graph-based DTI prediction.
• The proposed method advances computational strategies in drug discovery and development.