Skip to content

Parameter Estimation

This module provides the tools needed to estimate parameters for ODE‐based models of phosphorylation dynamics.

Overview

The module is organized into several submodules:

  • normest.py – Implements normal parameter estimation. This approach fits the entire time-series data in one step.
  • toggle.py – Offers a single function (estimate_parameters) to pipe normal estimation based on a mode flag.
  • core.py – Integrates the estimation methods, handling data extraction, calling the appropriate estimation (via the toggle), ODE solution, error calculation, and plotting.

Features

  • Bootstrapping:
    Bootstrapping can be enabled to assess the variability of the parameter estimates.

  • Flexible Model Configuration:
    The module supports different ODE model types (e.g., Distributive, Successive, Random) through configuration constants. For example, when using the "randmod" (Random model), the parameter bounds are log-transformed and the optimizer works in log-space (with conversion back to the original scale).

  • Integration with Plotting:
    After estimation, the module calls plotting functions (via the Plotter class) to visualize the ODE solution, parameter profiles, and goodness-of-fit metrics.

  • Networkmodel scalar path:
    The current networkmodel parameter-estimation path uses GlobalODEScalarObjective and jaxopt.ProjectedGradient with projection=project_bounds. The stopping criteria are JAXopt parameter-distance convergence, ‖x_k - x_{k-1}‖ < tol, or the hard iteration ceiling iter_num >= maxiter.