Reproduce “Calibration of stochastic, agent-based neuron growth models with Approximate Bayesian Computation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Tobias Duswald, Lukas Breitwieser, Thomas Thorne, Barbara Wohlmuth, Roman Bauer

A framework employing Approximate Bayesian Computation (ABC) is presented for calibrating stochastic, agent-based neuron growth models, inferring parameter posterior distributions and quantifying uncertainties. The approach, utilizing statistical distances on morphometric distributions, accurately recovered parameters for synthetic data and provided convergent calibrations for human and mouse hippocampal pyramidal cells.

Key insights

  • The choice and number of morphometrics are critical for accurate parameter inference; a comprehensive set significantly improved posterior concentration.
  • Wasserstein distance demonstrated superior reliability and convergence compared to other statistical distances (KL, "gamma"-divergence, Sliced-Wasserstein) for comparing distributions of neuronal morphometrics in the ABC framework.
  • Efficient calibration requires balancing the number of simulations per parameter with the total simulation budget, showing that a moderate number of simulations per parameter (e.g., 25) can yield good posterior quality within reasonable runtime.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.