This code accompanies the paper https://biorxiv.org/cgi/content/short/2020.07.31.231621v2. Documentation is available at https://lineageot.readthedocs.io/.
First, clone this repository with
git clone https://github.com/aforr/LineageOT
Dependencies are listed in conda_requirements.yml as well as in pip_requirements.txt; earlier versions of the packages may work but have not been tested. If you are using Anaconda, they can be installed with
conda env create -f conda_requirements.yml
Installation may take a few minutes. Activate the environment with
conda activate lineageOT
If you are using pip and not Anaconda, install dependencies with
pip install numpy cython
pip install -r pip_requirements.txt
Once you have set up and activated the environment, install the most recent LineageOT by running
pip install .
from the repository's base directory. Alternatively, install from PyPI with
pip install lineageot
No specific operating system is required, though there may be a bug in one of the dependencies in certain versions of MacOS (PythonOT/POT#93). The code has been tested on OS X 10.14.6 and Ubuntu 16.04.
An example of LineageOT applied to simulated data is in examples/simulation_demo.ipynb. Running the notebook for one simulation type took around 10 minutes on a "normal" desktop.
To fit LineageOT couplings in your own system, follow the steps in examples/pipeline_demo.ipynb replacing the synthetic AnnData object with your data.
We recommend then using the downstream analysis tools available in the Waddington-OT package: https://broadinstitute.github.io/wot/.