Drug-target interaction prediction based on spatial consistency constraint and graph convolutional autoencoder.

Area of Science:

• Computational biology
• Bioinformatics
• Drug discovery

Background:

• Drug-target interaction (DTI) prediction is crucial for drug discovery and repositioning.
• Existing computational methods often fail to preserve nearest neighbor relationships during representation learning, limiting DTI prediction performance.
• Topological relationships between drugs and targets are vital for accurate DTI prediction.

Purpose of the Study:

• To propose a novel graph convolutional autoencoder-based model, SDGAE, for enhanced DTI prediction.
• To address the limitations of existing methods by incorporating topological invariance in representation learning.
• To improve the accuracy and robustness of DTI prediction.

Main Methods:

• Developed SDGAE, a graph convolutional autoencoder model for DTI prediction.
• Implemented a pre-processing step to reduce isolated nodes in the heterogeneous network, enabling effective graph convolutional network utilization.
• Ensured graph structure maintenance during representation learning to preserve nearest neighbor relationships.

Main Results:

• SDGAE effectively predicts drug-target interactions by maintaining the graph structure.
• The model learns informative and robust feature vectors for drugs and targets.
• Significantly improved predictive accuracy for DTIs was achieved compared to existing methods.

Conclusions:

• SDGAE offers a powerful approach for DTI prediction by preserving topological relationships.
• The model's ability to learn robust features enhances its utility in drug discovery and repositioning.
• SDGAE demonstrates superior performance in predicting drug-target interactions.