HySimODE: a hybrid stochastic-deterministic simulation framework for multiscale models of biological systems.

Area of Science:

• Biochemistry and Systems Biology
• Computational Biology
• Biophysics

Background:

• Hybrid simulation is crucial for modeling biochemical systems with both low-copy stochastic and high-abundance deterministic dynamics.
• Existing methods often require manual partitioning of species into different dynamic regimes, which is labor-intensive and prone to error.

Purpose of the Study:

• To present HySimODE, an automated Python framework for hybrid simulation of biochemical systems defined by ordinary differential equations (ODEs).
• To enable robust and reproducible hybrid simulations by eliminating the need for manual regime specification.

Main Methods:

• HySimODE employs a machine-learning classifier, trained on diverse ODE models, to automatically assign species to stochastic or deterministic states.
• It integrates a simple stochastic update rule with a stiff ODE solver within a unified simulation loop.
• A modular adapter system converts concentration-based ODE models to molecular counts for universal compatibility.

Main Results:

• The machine-learning classifier achieves robust stochastic-deterministic partitioning beyond simple abundance thresholds.
• HySimODE successfully models systems with complex kinetics, including saturable reactions and effective-rate laws.
• Benchmarking confirms HySimODE's practical scalability and accuracy compared to traditional methods.

Conclusions:

• HySimODE offers a practical, scalable, and automated solution for hybrid simulations of ODE-defined biochemical models.
• The framework facilitates data-driven modeling and analysis of complex biological systems in synthetic biology and neurobiology.
• Automated partitioning and universal compatibility enhance the accessibility and reproducibility of hybrid simulations.