Physics Informed Neural Network based method for solving the single-cell population dynamics.
For each cell, we estiamte the dynamic parameter of the cell proliferation, differentiation and diffusion.
Check the tutorial notebooks for instructions on preparing your data and downstream analysis.
git clone https://github.com/Gottgens-lab/pseudodynamics_plus.git
cd pseudodynamics_plus
pip install -e .To train pseudodynamics+ on your data, make sure you store the population size information in AnnData.uns['pop'] and saved in h5ad format. Configure the training setting in config.json. See examples here example. Run the following command:
# with GPU
python main_train.py --config_path config.json -G 0
# without GPU
python main_train.py --config_path config.json -G None