BANG - BooleAn Networks on the GPU

BANG is a Python package dedicated to analysis, simulation, and control of Boolean networks with the help of CUDA.

Table of Contents

• Features
• Installation
• Usage
• Development

Features

• High-performance simulation: GPU-accelerated computations for Boolean network simulations.
• Effective block-based attractor detection (PBN): Efficiently find attractors in large networks.
• Visualization: Built-in support for visualizing network dynamics.
• Parsing: Easy-to-use parsing of PBNs from ASSA-PBN and SBML formats.

Installation

BANG is available on PyPI and can be installed using pip. Ensure you have Python 3.10 or higher.

pip install bang-gpu

For the latest development version, you can clone the repository and install it manually. First you need to clone the repository:

git clone https://github.com/zpp20/bang.git
cd bang

Now, you can install the package:

pip install .

Usage

1. Creating a Boolean Network

You can define a Boolean network using a set of parameters of a Probabilistic Boolean Network. For example:

from bang import PBN

pbn = PBN(
    # number of nodes
    3, 
    # number of functions
    [1, 1, 1],
    # number of parents in functions
    [2, 2, 1], 
    # functions
    [[True, True, True, False], [False, True, False, False], [True, False]],
    # indexes of the parents in the functions
    [[0, 1], [0, 1], [1]],
    # probabilities that the functions are chosen
    [[1.0], [1.0], [1.0]],
    # probability of a random perturbation at any step
    0.0,
    # initial state
    [3],
)

2. Simulating how the network evolves

Run a simulation for a specified number of steps:

# Set the initial state (otherwise the default is all False)
pbn.set_initial_state([[True, False, True]])

# Simulate the network for 10 steps
pbn.simple_steps(10)

# Access the simulation history
print("Trajectory history:", pbn.history)
# Access the final state
print("Last state:", pbn.last_state)

3. Using CPU-Based simulation

If you want to run the simulation on the CPU instead of the GPU:

# Simulate the network on the CPU for 10 steps
pbn.simple_steps(n_steps=10, device="cpu")

4. Detecting attractors

To detect attractors in the network, you can use the attractor_detection method:

# Detect attractors in the network using the block-based method
attractors = pbn.blocks_detect_attractors()
print("Attractors found:", attractors)

5. Visualization

You can visualize the state transitions of the network:

# Plot the evolution of the trajectory simulated in the 1st thread
pbn.trajectory_graph(1)

Development

Setup

The easiest way to set up all the necessary development dependencies is to use pip with the dev extras:

pip install -e ".[dev]"

Code quality

BANG uses black for code formatting and isort for import sorting. It also uses ruff To ensure code quality, you can run the following commands:

# Format the code
black .
isort .

Testing

BANG uses pytest for testing. To run the tests, you can use the following command:

pytest tests/

Linting

BANG uses ruff for linting. To run the linter, you can use the following command:

ruff check .

Release process

In order to release a new version of BANG, you need to follow these steps:

1. Update the version in pyproject.toml.

2. Commit the changes with a message like "Bump version to X.Y.Z".

3. Tag the commit with the new version number: git tag vX.Y.Z. Beware that the tag must be in the format vX.Y.Z[-prerelease] (e.g., v1.0.0).

4. Push the changes and the tag to the remote repository: git push origin main --tags.

• If the tag indicates a pre-release version and is not on the main branch, CD will deploy the package to TestPyPI.

• If the commit is on main and the tag does not contain a prerelease version, it will be deployed to PyPI.