fedRBE

Local test simulation

This guide is designed for people who want to test the fedRBE tool locally on their machine via command line.

For more detailed information and advanced usage, please refer to the Documentation.

Prerequisites and setup

Ensure you have the following installed before starting:

1. Docker: Installation Instructions
2. Python 3.8+: Installation Instructions
3. FeatureCloud CLI:

   pip install featurecloud
   

4. App Image (either pull or build locally — see Installation).

For Windows users, git must also be installed and added to PATH. We recommend using WSL.

Usage

Run simulations locally to understand fedRBE’s behavior:

1. Ensure the full repository including sample data is cloned and the current working directory:

   git clone https://github.com/Freddsle/fedRBE.git
   cd fedRBE
   

   Note: Some of the sample data used in this guide is tracked by Git LFS. If you need the full evaluation datasets, install Git LFS and run git lfs pull after cloning.

2. Start the FeatureCloud Controller with the correct input folder:

   featurecloud controller start --data-dir=./evaluation_data/simulated/mild_imbalanced/before/
   

3. Run a Sample Experiment:

   featurecloud test start --app-image=featurecloud.ai/bcorrect:latest --client-dirs=lab1,lab2,lab3
   

   Alternatively, you can start the experiment from the frontend.
   Select 3 clients, add lab1, lab2, lab3 respecitvely for the 3 clients to their path.

   Use featurecloud.ai/bcorrect:latest as the app image.

   You can monitor the logs of the batch effect correction run and receive the final results of the different clients on the frontend in your browser.

For a step-by-step detailed instructions on how to start collaboration using multiple machines, refer to the How To Guide.

This site is open source. Improve this page.