sebapersson/sysbionotebooks

SysBioNotebooks

Example Jupyter Notebooks for working with PEtab.jl and SBMLImporter.jl

This repository contains example Jupyter Notebooks for working with the Julia packages SBMLImporter.jl and PEtab.jl.

Installation

To set up the notebooks, first clone the repository:

git clone https://github.com/sebapersson/SysBioNotebooks.git

Next, navigate into the cloned repository, start Julia, and in the Julia REPL enter:

] instantiate

or alternatively

import Pkg; Pkg.instantiate

For best performance, we strongly recommend using the latest version of Julia, which currently is Julia 1.10.

SBMLImporter.jl examples

SBMLImporter.jl is a Julia importer for dynamic models defined in the Systems Biology Markup Language (SBML). It supports most SBML features such as events and dynamic compartment sizes. Imported models are converted into a Catalyst ReactionSystem. This ReactionSystem can then be transformed into various problem types, such as a JumpProblem for Gillespie simulations, a SDEProblem for Langevin SDE simulations, or an ODEProblem for deterministic simulations.

In the SBMLImporter folder of this repository, you'll find notebooks covering a range of topics:

• Importing a SBML model as a JumpProblem for Gillespie simulations, and subsequently evaluate the performance of different stochastic simulators on this problem.
• Importing a SBML model as an ODEProblem, and subsequently evaluate the performance of different ODE solvers on this problem. A Google Colab of this notebook can also be found here.

PEtab.jl examples

PEtab.jl is a Julia package for constructing parameter estimation or Bayesian inference problems for ODE models. It allows for the direct importation of problems formatted in the PEtab standard, alternatively the problem can be directly coded in Julia where the dynamic model can be provided either as a ModelingToolkit.jl ODESystem or a Catalyst ReactionSystem.

In the PEtab directory of this repository, you'll find notebooks covering a range of topics:

• Setting up a PEtab problem in Julia. A Google Colab of this notebook can also be found here
• Importing and simulating the ODE dynamic models in a PEtab problem.
• Benchmarking different ODE solvers for a PEtab problem.
• Benchmarking different gradient computation methods, such as forward and adjoint methods, for a PEtab problem.
• Performing parameter estimation for a PEtab problem.
• Performing Bayesian inference for a PEtab problem, using either Adaptive Metropolis-Hastings methods or Hamiltonian Monte Carlo techniques (such as NUTS, which is used in Turing.jl).