REPOP is a probabilistic modeling library designed to reconstruct bacterial population distributions from plate counting data. It provides efficient inference methods with built-in uncertainty quantification.
A more detailed explanation is available in our manuscript:
REPOP: A Tool for Bacterial Population Reconstruction with Uncertainty Quantification from Plate Counts
Available on eLife
REPOP is optimized for ease of use and computational efficiency. While it has minimal dependencies, it benefits significantly from GPU acceleration.
To install REPOP, simply clone the repository (or download the files, extract and put on folder) and run:
pip install .REPOP requires only a few standard scientific computing libraries. These are installed automatically using the command above.
torch,numpy,matplotlib,scikit-learn
To get started, check out the Jupyter notebook Tutorial.ipynb, which demonstrates how to use REPOP with plate counting data.
To generate synthetic data used in the manuscript, navigate to the synth_data folder and run:
python cases.pyIf you find our software or research helpful in your work, please consider citing our paper:
@article {Pessoa2025repop,
author = {Pessoa, Pedro and Lu, Carol and Tashev, Stanimir Asenov and Kruithoff, Rory and Shepherd, Douglas P. and Press{\'e}, Steve},
title = {REPOP: bacterial population quantification from plate counts},
journal = {eLife},
publisher = {eLife Sciences Publications, Ltd},
year = {2025},
doi = {10.7554/elife.107122.1}
}
The graphs for Fig. 1 are generated when running the tutorial. All others can be generated by the respective python script with its number in the file's name.