Squeeze-and-breathe evolutionary Monte Carlo optimization with local search acceleration and its application to...

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• Parameter estimation is crucial for biological modeling, but often challenged by sparse and noisy datasets.
• Existing heuristic methods have limitations in biological applications, especially when parameter ranges are unknown.

Purpose of the Study:

• To develop a robust and efficient algorithm for model parameter fitting in biological systems.
• To address challenges posed by unknown parameter magnitudes and ranges in complex datasets.

Main Methods:

• A novel algorithm combining evolutionary algorithms, sequential Monte Carlo, and direct search optimization.
• Iterative refinement of parameter distributions using local optimization and partial resampling.
• Utilizes a historical prior defined over all previous iterations for robust estimation.

Main Results:

• The algorithm performs well even with unknown parameter orders of magnitude or ranges.
• Efficient parameter estimation demonstrated on both simulated and real biological experimental data.
• Successfully estimates parameters in the absence of a priori knowledge.

Conclusions:

• The developed algorithm offers an efficient and robust solution for parameter estimation in biological modeling.
• Applicable to diverse biological systems, enhancing the reliability of model-based analyses.
• Provides a valuable tool for researchers dealing with complex, data-limited biological models.