KGRLFF: Detecting Drug-Drug Interactions Based on Knowledge Graph Representation Learning and Feature Fusion

Area of Science:

• Pharmacology
• Bioinformatics
• Computational Chemistry

Background:

• Accurate drug-drug interaction (DDI) prediction is crucial for drug development and patient safety.
• Existing DDI prediction models often rely on single data sources and struggle with biomedical knowledge graphs.

Purpose of the Study:

• To propose a novel hybrid method, KGRLFF, for enhanced DDI prediction.
• To leverage both biomedical knowledge graphs and drug molecular structures for improved prediction.

Main Methods:

• Utilized Bidirectional Random Walk sampling (BRWP) for higher-order neighborhood information from knowledge graphs.
• Employed Knowledge Graph-based Cyclic Recursive Aggregation (KGCRA) for learning drug embeddings.
• Integrated drug molecular structures for structured feature learning.
• Developed a Feature Representation Fusion Strategy (FRFS) to combine embeddings and structured features.

Main Results:

• The KGRLFF method demonstrated feasibility in predicting potential drug-drug interactions.
• The hybrid approach effectively exploits information from both knowledge graphs and molecular structures.

Conclusions:

• KGRLFF offers a promising approach for accurate DDI prediction.
• Integrating diverse data sources enhances the performance of DDI prediction models.