A cooperative strategy for parameter estimation in large scale systems biology models.

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• Mathematical models are crucial in systems biology for knowledge summarization and prediction.
• Model calibration, or parameter estimation, is challenging due to non-linearity, numerous parameters, and scarce data.
• Efficient global optimization methods are needed for accurate biological model calibration.

Purpose of the Study:

• To introduce a novel, efficient global optimization method for parameter estimation in large-scale biological models.
• To address the challenges of non-linearity, high dimensionality, and data scarcity in systems biology model calibration.

Main Methods:

• Cooperative Enhanced Scatter Search (CeSS), a parallel computing approach utilizing multiple cooperating threads.
• Each thread employs the enhanced Scatter Search (eSS) metaheuristic.
• Information sharing between parallel threads to enhance systemic algorithm properties and performance.

Main Results:

• CeSS demonstrated superior performance in parameter estimation for large-scale models, including those of E. coli central carbon metabolism.
• The method was validated on benchmark global optimization problems, yielding excellent results.
• Cooperation between threads significantly speeds up the optimization process.

Conclusions:

• CeSS is a versatile technique applicable to diverse model calibration problems in systems biology.
• The method outperforms existing approaches for large-scale model calibration.
• CeSS is extensible, allowing integration with other solvers and problem-specific information.