scdynsys

Methods for fitting dynamical models to single-cell data

Our paper about this method and application to resident-memory T cell data has been published in PLOS Computational Biology:

CH van Dorp, JI Gray, DH Paik, DL Farber, AJ Yates. A Variational deep-learning approach to modeling memory T cell dynamics PLoS Comput Biol 21(7): e1013242 (2025)

Installation and testing

To install the scdynsys package, first clone the git repository, and then create a virtual environment

git clone git@github.com:chvandorp/scdynsys.git
cd scdynsys
python3 -m venv .venv

Here it is assumed that python3 points to a python version >= 3.10. Next, activate the virtual environment and install the package. This will download and install all dependencies.

source .venv/bin/activate
pip install .

You can use pytest to see if something isn't working. This will collect all unittests in the tests folder and run them.

pip install pytest
python3 -m pytest

Downloading the data files

Large data files are hosted on Zenodo. To run the notebooks (see below), you first have to download the flow cytometry data from Zenodo. Downloading the files can be done with the fetch_data command (provided by the scdynsys package):

fetch_data

The files are download to the data folder.

Using the package

The notebooks folder contains code used for the aforementioned preprint. this folder also constans an overview (README) file describing what all notebooks are used for and the order in which they have to be executed.

TODO

Some improvements:

• Replace python pickles in the data folder with human-readable json files.
• Refactor the notebooks to avoid code repetition between the CD8 and CD4 analysis.