Computational Advancements in Cancer Combination Therapy Prediction

Area of Science:

• Computational biology
• Pharmacology
• Oncology

Background:

• High attrition rates in de novo drug discovery necessitate cost-effective alternatives.
• Single-agent cancer therapies often exhibit limited efficacy.
• In silico drug repurposing offers a promising strategy for identifying novel combination therapies.

Purpose of the Study:

• To introduce computational methods for predicting cancer combination therapies.
• To summarize recent studies utilizing these prediction methods.
• To highlight considerations for improving future prediction models.

Main Methods:

• Systematic literature search of PubMed (last 10 years).
• Inclusion of reviews and articles on ongoing/retrospective studies.
• Focus on methods for improving combination therapy prediction.

Main Results:

• Computational methods include network analysis, regression-based machine learning, classifier machine learning, and deep learning.
• Each method class presents unique advantages and disadvantages.
• Prioritization of studies offering insights for method enhancement.

Conclusions:

• Careful selection of computational methods is crucial for designing effective prediction strategies.
• Future improvements involve integrating disease pathobiology, drug characteristics, multiomics data, and drug-drug interactions.
• Advanced computational integration of diverse data will enable more accurate prediction of safe and efficacious cancer combination therapies.