A Python package for the mathematical modeling of infectious diseases via compartmental models. Originally designed for epidemiologists, epispot can be adapted for almost any type of modeling scenario.
The epispot package currently only supports compartmental models, though we plan to expand the package to work for stochastic agent-based and spatial models as well. Currently, epispot offers the following:
- Quick compilation of compartmental models with the following compartments:
- Custom-defined compartments for research
- Built-in graphing and visualization engine
- Plots model predictions interactively
- Creates comparisons between models
Due to its diverse range of features, epispot can be used for both research and experimental modeling. If you would like to add more modeling support, please see the contributing section.
The epispot package can be installed from PyPI, Anaconda, or be built from the source. Before reading this guide, it is important to note that there are actually two different versions of the epispot package. The first of which is the
master package, which will always have a version tag like
v#.#.#. This package is used to release stable versions of epispot. However, during important events, like the COVID-19 pandemic, the
nightly package is used to publish new features quickly. However, these versions may be unstable.
This is the easiest way to install epispot. Fire up a terminal and type:
pip install epispot
For the nightly version, use
pip install epispot-nightly
Pip will ask you to install
matplotlib as dependencies if you haven't already. Additionally, it may require you to install
fire for the CLI.
These can be installed beforehand with:
pip install numpy pip install matplotlib pip install fire
Please note that the
nightly version is not available on the
conda package registry. However, it is still possible to install on
conda-based systems with
pip install epispot-nightly
pip from Anaconda to install it.
The standard version of epispot is published to
conda using the
conda-forge channel. To install, please use:
conda config --add channels conda-forge conda install -c conda-forge epispot
Building from the source
This is the hardest way to install
epispot and it is recommended that you use either
Anaconda to install it instead. However, if you would like to contribute to the repository, this will be particularly useful.
Clone the repository with:
git clone https://github.com/epispot/epispot # clone epispot/epispot cd epispot # open project pip install -r requirements.txt # install package requirements pip install -r bin/requirements.txt # Install CLI requirements
Then, build the nightly version with
python setup-nightly.py install
For the stable version, first checkout a release branch with something like:
git checkout v2.1.1 python setup.py install
Using epispot in a Python REPL to create the well-known SIR model (in less than three lines of code):
The documentation for the epispot package is generated automatically from the Python source code using Pdoc3. You can view the documentation for both the nightly and stable builds of epispot here.
At first, the documentation may seem a bit hard to understand, especially if you're new to epidemiology. That's why epispot has put together an entire manual describing some basic concepts you'll need to know to master epispot. You can view it here. The GitHub source is available here.
The GitHub repository has a vast array of samples using epispot. You can start by checking out the
explorables/ directory. In it, you'll find many programs designed for helping you get started with epispot and some hands-on examples.
If you have any feedback, please
- Create a discussion on GitHub
- Create an issue if you've found a bug
- Submit a PR if you want to add a new feature
- Contact a CODEOWNER
Contributions are always welcome! See CONTRIBUTING.md for instructions on how to get started, including environment setup and instructions to build from the source. Please note also that epispot has many guides dedicated to certain types of contributions. Please see
If you plan on using epispot in your project, please abide by the GPLv3 license. This requires that any changes you make to epispot must be open-sourced under the GPLv3 license as well and that you give credit to the author, which you can do by citing the project in your research, linking back to the original repository, or mentioning the author @quantum9innovation.
For research, you can also use epispot's DOI to reference the project:
The recommended citation for epispot is:
quantum9innovation (2021, April 2). epispot/epispot: (Version 2.1.0). Zenodo. http://doi.org/10.5281/zenodo.4624423
There are many related projects to epispot, although we believe that epispot has greater extensibility than many of these other projects. Additionally, epispot is quite portable (graphs can be used as web components/displayed as images/etc.). However, below we would like to acknowledge a few projects that may be better suited to certain use cases:
- covasim by the Institute for Disease Modeling offers agent-based stochastic models which epispot unfortunately does not support as of yet
- CovsirPhy by @lisphilar offers greater support for loading and analyzing real COVID-19 data, which epispot strives to add soon
Please see our CODEOWNERS file for authors. Because epispot is an open-source project, different pieces of our code have different authors. However, if citing epispot or using it in another project, you can put @quantum9innovation as the lead author.
Idea & Inspiration
The original idea for epispot came from a 3Blue1Brown video on basic infectious disease dynamics and an interactive article in the Washington Post. This in turn inspired the very basic infectious disease dynamics simulated here. However, what finally set the package into motion was a series of articles by Henry Froese, available on Medium here, along with their corresponding interactive notebooks.
Code Development & Maintenance
The epispot project is built on open source code and is itself open-source. The initial core development was fueled by @quantum9innovation and much of the codebase was maintained by @Quantalabs. Additionally, thank you to all of epispot's open-source contributors!
The epispot team also relies on the following open-source projects as dependencies:
- NumPy (GitHub), the fundamental package for scientific computing with Python
- Matplotlib (GitHub), plotting with Python
- Google Fire (Github), a library for automatically generating command line interfaces (CLIs) from absolutely any Python object.
External Code Management Tools
For code maintenance, epispot uses various tools including:
- Coverage.py (PyPI) for code coverage report generation
- Pdoc3 (GitHub) for automatic documentation generation
- GitBook (Website) for documentation hosting
- CodeCov (Website) for code coverage report analysis
- LGTM (Website) for CodeQL analysis
- DeepSource (Website) for static code analysis
- GitLocalize (Website) for localization of documentation
- Zenodo (Website) for automatic DOI & citations