FluxDisco: Symbolic Regression for Stoichiometric Dynamical Systems via Monte Carlo Graph Search

FluxDisco is a physical information framework for ODE systems based on flux and stoichiometry. This method utilizes a Monte Carlo graph search algorithm to identify governing differential equations that conform to physical laws from noisy data. By introducing known stoichiometric relationships to reduce the expression search space, this method ensures the physical consistency of the recovered results. Currently, FluxDisco has been validated in various physical and biological systems and demonstrates the ability to accurately recover interpretable dynamics.