BluePyOpt: Leveraging Open Source Software and Cloud Infrastructure to Optimise Model Parameters in Neuroscience.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Mathematical models in neuroscience often involve complex dynamical systems.
• Parameterizing these models requires solving global non-linear optimization problems with challenging fitness landscapes.
• Stochastic optimization, like evolutionary algorithms, is effective but demands significant expertise to set up and configure.

Purpose of the Study:

• To introduce BluePyOpt, a Python package designed to simplify data-driven parameter optimization for neuroscience models.
• To provide an extensible framework that standardizes and integrates existing open-source tools for model fitting.
• To facilitate the creation and sharing of optimization techniques and knowledge within the neuroscience community.

Main Methods:

• BluePyOpt abstracts optimization and evaluation tasks into reusable, flexible elements based on best practices.
• The package wraps and standardizes several existing open-source tools for parameter optimization.
• It offers methods for setting up optimizations on diverse platforms, from laptops to clusters and cloud infrastructures.

Main Results:

• BluePyOpt successfully simplifies the complex task of parameterizing neuroscience models.
• The framework enables efficient data-driven model optimization by abstracting technical complexities.
• Demonstrated versatility through three representative neuroscience use cases.

Conclusions:

• BluePyOpt lowers the barrier to entry for advanced computational modeling in neuroscience.
• The package promotes reproducibility and knowledge sharing in the field.
• It empowers researchers to perform sophisticated model parameter optimization across various computational environments.