This is a simple CLI and a library for BioNetGen modeling language. PyBioNetGen also includes a heavily updated version of Atomizer which allows for conversion of models written in Systems Biology Markup Language (SBML) into BioNetGen language (BNGL) format.
Please see the documentation to learn how to use PyBioNetGen.
You will need both python (3.7 and above) and perl installed. Once both are available you can use the following pip command to install the package
$ pip install bionetgen
PyBioNetGen can optionally use BNGsim as an in-process simulation engine (ODE/SSA/NFsim, multi-format input). This is fully opt-in: if BNGsim is not installed, every simulation uses the existing subprocess BNG2.pl path and behavior is unchanged. BNGsim is never a required dependency.
BNGsim is a compiled extension and is not on PyPI yet, so pip install bionetgen[bngsim] will not resolve until it is published. To try it now, install a prebuilt wheel from the release assets (macOS / Python 3.12; build from source on other platforms). PyBioNetGen requires BNGsim 0.9.10 or newer. Set BIONETGEN_NO_BNGSIM=1 to force the legacy subprocess path even when BNGsim is installed. See the BNGsim documentation page for details.
PyBioNetGen comes with a command line interface (CLI), based on cement framework, as well as a functional library that can be imported. The CLI can be used to run BNGL models, generate Jupyter notebooks and do rudimentary plotting.
The library side provides a simple BNGL model runner as well as a model object that can be manipulated and used to get libRoadRunner simulators for the model.
PyBioNetGen also includes a heavily updated version of Atomizer which allows for conversion of SBML models into BNGL format. Atomizer can also be used to automatically try to infer the internal structure of SBML species during the conversion, see here for more information. Please note that this version of Atomizer is the main supported version and the version distributed with BioNetGen will eventually be deprecated.
The model object requires a system call to BioNetGen so the initialization can be relatively costly, in case you would like to use it for parallel applications, use the libRoadRunner simulator instead, unless you are doing NFSim simulations.
Sample CLI usage
$ bionetgen -h # help on every subcommand
$ bionetgen run -h # help on run subcommand
$ bionetgen run -i mymodel.bngl -o output_folder # this runs the model in output_folder
Sample library usage
import bionetgen ret = bionetgen.run("/path/to/mymodel.bngl", out="/path/to/output/folder")
# out keyword is optional, if not given, # generated files will be deleted after running
res = ret.results['mymodel']
# res will be a numpy record array of your gdat results
model = bionetgen.bngmodel("/path/to/mymodel.bngl")
# model will be a python object that contains all model information
print(model.parameters) # this will print only the parameter block in BNGL format
print(model) # this will print the entire BNGL
model.parameters.k = 1 # setting parameter k to 1
with open("new_model.bngl", "w") as f:
f.write(str(model)) # writes the changed model to new_model file
# this will give you a libRoadRunner instance of the model
librr_sim = model.setup_simulator()
You can find more tutorials here.
The following demonstrates setting up and working with a development environment:
### create a virtualenv for development
$ make virtualenv
$ source env/bin/activate
### run bionetgen cli application
$ bionetgen --help
### run pytest / coverage
$ make test
Included is a basic Dockerfile for building and distributing BioNetGen CLI,
and can be built with the included make helper:
$ make docker
$ docker run -it bionetgen --help
You can use make dist command to make the distribution and push to PyPI with
python -m twine upload dist/*
You'll need to have a PyPI API token created, see here for more information.