A diverse stochastic search algorithm for combination therapeutics

Area of Science:

• Computational biology
• Pharmacology
• Bioinformatics

Background:

• Designing effective drug combination cocktails for individual patients is challenging due to the vast number of potential combinations and patient-specific genetic variations.
• Exhaustive experimental approaches are infeasible, and existing methods struggle to incorporate personalized genetic data for optimization.

Purpose of the Study:

• To develop an efficient algorithm for optimizing drug combination cocktails.
• To minimize the number of experiments required for identifying effective drug combinations for personalized medicine.

Main Methods:

• A novel stochastic search algorithm combining parallel experimentation with focused and diversified sequential search.
• Evaluation on synthetic datasets and biological examples, including bacterial and lung cancer cell inhibition responses.

Main Results:

• The proposed algorithm demonstrated superior performance compared to a recently proposed adaptive reference update approach.
• Achieved an average reduction of 45% in experimental iterations across all tested examples.
• Successfully optimized drug combinations for bacterial and lung cancer cell inhibition.

Conclusions:

• The diverse stochastic search algorithm efficiently identifies optimized drug combinations in fewer iterative steps.
• This approach is amenable to integration with patient genetic data for designing personalized drug cocktails.
• Offers a promising strategy for accelerating drug discovery and development in personalized medicine.