GeNNius: an ultrafast drug-target interaction inference method based on graph neural networks

Area of Science:

• Computational biology
• Bioinformatics
• Drug discovery

Background:

• Drug-target interaction (DTI) prediction is crucial for drug repurposing but faces challenges with existing computationally expensive methods.
• Current DTI prediction approaches lack generalization capabilities, hindering robust development.

Purpose of the Study:

• To introduce GeNNius, a Graph Neural Network (GNN)-based system for accurate and efficient DTI prediction.
• To evaluate GeNNius's performance, prediction power for novel interactions, and generalization capabilities across diverse datasets.

Main Methods:

• Developed GeNNius, a Graph Neural Network (GNN)-based method for DTI prediction.
• Assessed accuracy, time efficiency, and prediction of novel DTIs on multiple datasets.
• Evaluated generalization by training and testing on different datasets.

Main Results:

• GeNNius outperforms state-of-the-art models in accuracy and time efficiency.
• Demonstrated prediction of previously unknown DTIs.
• Showcased strong generalization capabilities, enabling training on large datasets and testing on smaller ones.

Conclusions:

• GeNNius offers a computationally efficient and accurate solution for DTI prediction.
• The GNN encoder preserves biological information, distinguishing protein families in embeddings.
• GeNNius has the potential to significantly improve drug repurposing through enhanced DTI inference.