Synergizing Anti-Cancer Drug Combinations With Dual-View Hypergraph Representation Fusion.

Area of Science:

• Computational biology
• Drug discovery
• Bioinformatics

Background:

• Drug combination therapy is crucial for treating diseases like cancer, enhancing efficacy and overcoming drug resistance.
• Identifying synergistic drug combinations is challenging due to the vast combinatorial space and limitations in existing methods.
• Current approaches often fail to fully utilize the complex relationships within known synergistic combinations.

Purpose of the Study:

• To propose a novel deep learning model, DVHSyn, for accurate identification of synergistic drug combinations.
• To leverage dual-view hypergraph representation fusion to capture both local and global context of drug-target interactions.
• To improve the prediction of synergistic drug combinations for enhanced cancer treatment and drug development.

Main Methods:

• DVHSyn extracts transcriptome features from cancer cell lines and molecular structures from drugs.
• It models synergistic effects using a hypergraph, learning from both hypergraph and expanded heterogeneous graph views.
• A selective fusion of learned representations from dual views predicts synergistic drug combinations.

Main Results:

• DVHSyn outperformed six existing methods in identifying synergistic drug combinations.
• Experimental results validate the model's effectiveness and potential for predicting novel synergistic drug pairs.
• A case study demonstrated DVHSyn's capability in discovering new synergistic combinations.

Conclusions:

• DVHSyn offers an effective deep learning approach for synergistic drug combination identification.
• The model provides new insights for developing novel drug therapies, particularly in oncology.
• This method enhances the prediction of synergistic drug combinations, aiding future drug discovery efforts.