NEXGB: A Network Embedding Framework for Anticancer Drug Combination Prediction.

Area of Science:

• Computational biology
• Bioinformatics
• Cancer research

Background:

• Drug combinations offer superior cancer treatment potential compared to single agents.
• Current prediction methods primarily use genomic and chemical data, neglecting protein-protein interaction (PPI) network topology.
• Integrating PPI network information is crucial for enhancing drug combination synergy prediction.

Purpose of the Study:

• To develop a novel network-embedding-based prediction model, NEXGB, for identifying synergistic drug combinations against cancer cell lines.
• To incorporate topological features from PPI networks into drug and cancer cell line representations.
• To improve the accuracy of predicting drug combination efficacy in cancer therapy.

Main Methods:

• Utilized struc2vec to extract topological features of protein nodes within PPI networks.
• Integrated extracted topological features with drug target protein information to generate comprehensive drug and cancer cell line features.
• Employed extreme gradient boosting (XGBoost) for predicting synergistic relationships between drug combinations and cancer cell lines.
• Validated the model on the Oncology-Screen and DrugCombDB datasets.

Main Results:

• NEXGB demonstrated superior performance compared to five existing methods on two independent datasets.
• The model significantly improved the predictive power for discovering synergistic drug combinations and their relationships with cancer cell lines.
• Topological information from PPI networks was shown to be a valuable component for predicting effective combination therapies.

Conclusions:

• The NEXGB model effectively leverages PPI network topology for enhanced prediction of cancer drug combination synergy.
• Integrating network information provides a promising avenue for discovering novel combination therapies for cancer and other complex diseases.
• NEXGB offers a robust computational approach to advance precision oncology and drug discovery efforts.