cayenne : Python package for stochastic simulations

Introduction

cayenne is a Python package for stochastic simulations. It offers a simple API to define models, perform stochastic simulations with them and visualize the results in a convenient manner.

Currently under active development in the develop branch.

Install

Install with pip:

$ pip install cayenne

Documentation

General: https://cayenne.readthedocs.io.
Benchmark repository, comparing cayenne with other stochastic simulation packages: https://github.com/Heuro-labs/cayenne-benchmarks

Usage

A short summary follows, but a more detailed tutorial can be found here. You can define a model as a Python string (or a text file, see docs). The format of this string is loosely based on the excellent antimony library, which is used behind the scenes by cayenne.

from cayenne.simulation import Simulation
model_str = """
        const compartment comp1;
        comp1 = 1.0; # volume of compartment

        r1: A => B; k1;
        r2: B => C; k2;

        k1 = 0.11;
        k2 = 0.1;
        chem_flag = false;

        A = 100;
        B = 0;
        C = 0;
    """
sim = Simulation.load_model(model_str, "ModelString")
# Run the simulation
sim.simulate(max_t=40, max_iter=1000, n_rep=10)
sim.plot()

Change simulation algorithm

You can change the algorithm used to perform the simulation by changing the algorithm parameter (one of "direct", "tau_leaping" or "tau_adaptive")

sim.simulate(max_t=150, max_iter=1000, n_rep=10, algorithm="tau_leaping")

Our benchmarks are summarized below, and show direct to be a good starting point. tau_leaping offers greater speed but needs specification and tuning of the tau hyperparameter. The tau_adaptive is less accurate and a work in progress.

Run simulations in parallel

You can run the simulations on multiple cores by specifying the n_procs parameter

sim.simulate(max_t=150, max_iter=1000, n_rep=10, n_procs=4)

Accessing simulation results

You can access all the results or the results for a specific list of species

# Get all the results
results = sim.results
# Get results only for one or more species
results.get_species(["A", "C"])

You can also access the final states of all the simulation runs by

# Get results at the simulation endpoints
final_times, final_states = results.final

Additionally, you can access the state a particular time point of interest $t$. cayenne will interpolate the value from nearby time points to give an accurate estimate.

# Get results at timepoint "t"
t = 10.0
states = results.get_state(t) # returns a list of numpy arrays

Benchmarks

	direct	tau_leaping	tau_adaptive
cayenne	✔️ Most accurate yet	✔️ Very fast but may need manual tuning	Less accurate than GillespieSSA's version
Tellurium	❗ Inaccurate for 2nd order	N/A	N/A
GillespieSSA	Very slow	❗ Inaccurate for initial zero counts	❗ Inaccurate for initial zero counts
BioSimulator.jl	❗ Inaccurate interpolation	❗ Inaccurate for initial zero counts	❗ Inaccurate for initial zero counts