GNNSynergy: A Multi-View Graph Neural Network for Predicting Anti-Cancer Drug Synergy.

Area of Science:

• Computational biology
• Pharmacology
• Bioinformatics

Background:

• Drug combinations are vital in cancer therapy for enhanced efficacy and overcoming resistance.
• The vast combinatorial space makes experimental screening of all drug combinations impractical.
• Accurate computational prediction of drug combinations is essential for guiding experimental efforts.

Purpose of the Study:

• To propose GNNSynergy, a novel computational method for predicting synergistic drug combinations.
• To develop a multi-view graph neural network framework for learning drug embeddings.
• To improve the accuracy of identifying effective drug combinations in cancer therapy.

Main Methods:

• Utilized a multi-view graph neural network (GNN) framework considering cell lines as main and sub-views.
• Constructed graphs representing drug synergistic and antagonistic interactions within each view.
• Employed an attention mechanism to integrate multi-view drug embeddings for synergy prediction.

Main Results:

• GNNSynergy demonstrated superior performance in predicting novel synergistic drug combinations.
• The method significantly outperformed existing state-of-the-art approaches.
• Extensive experiments were conducted using the DrugComb database.

Conclusions:

• GNNSynergy offers an accurate and efficient computational approach for drug synergy prediction.
• The multi-view GNN framework effectively learns drug embeddings for improved therapeutic predictions.
• This method can guide experimental screening, accelerating the discovery of effective cancer drug combinations.