A review of machine learning approaches for drug synergy prediction in cancer

Area of Science:

• Computational biology
• Pharmacology
• Bioinformatics

Background:

• Combinational pharmacotherapy demonstrates synergistic effects beneficial for treating complex diseases, including malignancies.
• Identifying effective drug combinations experimentally is often costly, infeasible, and challenging due to the vast number of potential combinations.
• Computational methods, particularly machine learning algorithms (MLAs), have emerged as powerful tools in pharmacology and bioinformatics.

Purpose of the Study:

• To review and present machine learning models developed for predicting synergistic drug combinations in cancer.
• To highlight the integration of diverse data types for enhancing prediction accuracy.
• To provide a comprehensive overview of computational strategies for prioritizing cancer pharmacotherapies.

Main Methods:

• Utilizing machine learning algorithms (MLAs) for predicting synergistic drug combinations.
• Integrating various data sources, including gene expression, protein-protein interactions, metabolite interactions, and pathway information.
• Incorporating pharmaceutical data such as chemical structure, molecular descriptors, and drug-target interactions.

Main Results:

• MLAs can effectively prioritize potential synergistic drug combinations for cancer treatment.
• The integration of multi-omics and pharmaceutical data significantly improves the accuracy of synergistic effect prediction.
• Computational approaches offer a feasible and cost-effective alternative to extensive experimental screening.

Conclusions:

• Machine learning algorithms are valuable tools for identifying synergistic drug combinations in oncology.
• Leveraging diverse biological and chemical data enhances the predictive power of computational models.
• This review provides a foundation for developing advanced computational strategies in cancer pharmacotherapy.