Prediction of Drug-Target Interactions Based on Hypergraph Neural Networks With Multimodal Feature Fusion.

Area of Science:

• Computational Biology
• Bioinformatics
• Drug Discovery

Background:

• Accurate drug-target interaction (DTI) prediction is crucial for efficient drug discovery and optimization.
• Traditional experimental methods for DTI prediction are often time-consuming and expensive.
• Existing computational models frequently rely on limited graph representations and fixed network structures.

Purpose of the Study:

• To introduce HyperGCN-DTI, a novel framework utilizing hypergraph neural networks and multimodal feature fusion for advanced DTI prediction.
• To capture high-order multi-entity relationships and local topological connections through a dual-channel architecture.
• To enhance the accuracy and robustness of DTI prediction, especially in sparse or noisy real-world datasets.

Main Methods:

• Developed a novel framework, HyperGCN-DTI, employing hypergraph neural networks.
• Integrated multimodal fused features, including pretrained language model embeddings and diverse biological networks.
• Implemented a dual-channel architecture to capture both local and higher-order structural dependencies.

Main Results:

• HyperGCN-DTI significantly outperformed state-of-the-art DTI prediction models across multiple datasets.
• The model demonstrated robustness on imbalanced and large-scale real-world datasets.
• Validation through biomedical literature and molecular docking confirmed the reliability of top-ranked predictions.

Conclusions:

• HyperGCN-DTI represents a significant advancement in DTI prediction by integrating diverse information sources with hypergraph representation.
• The framework offers enhanced accuracy and robustness, proving particularly effective in challenging data settings.
• HyperGCN-DTI provides a powerful and generalizable tool for accelerating drug development and target identification.