#27 · Systems biology

Stability and multistationarity in mechanistic models of biochemical regulation

The dominant thread is formal analysis of dynamical models whose structure is specified first and then interrogated for stability, multistationarity, identifiability, or approximation validity. Two items address reaction-network theory: an ALT-graph and splitting algorithm for next-generation matrix stability analysis of biochemical reactions, and a Latin hypercube and adaptive sampling framework with surrogate models and CN2 rule induction that approximates parameter regions for multistationarity and multistability in mass-action systems. A moment-closure study assesses Gaussian, log-normal, and negative-binomial closures for bursty gene expression with nonlinear degradation, examining both stability and accuracy. Related modeling work examines whether simple compartmental models can reproduce multiple epidemic waves, and a reduced ODE model of aquatic decomposition tests identifiability with synthetic experiments. What is new is explicit assessment of approximation validity and parameter-region description, rather than model fitting alone.