FL-Bioinformatics: Code Examples for Federated Learning

This repository contains code examples related to the review article "Technical Insights and Legal Considerations for Advancing Federated Learning in Bioinformatics," currently under evaluation for publication in OUP Bioinformatics.

We provide three different examples:

• WeaklyExpressed: Detecting weakly expressed genes in a federated dataset.
• MinorAllele: Computing minor allele frequencies in a federated dataset.
• SecureSum: Implementing a custom strategy in Flower for secure summation.

We recommend exploring the examples in the given order, as this will simplify the learning process. All examples refer the paper’s section "Practical insights on federation".

Please note that Flower does not provide a built-in strategy for federated secure summation (which is frequently mentioned in the article). Instead, it provides FedAvg as a built-in strategy, as it is far more commonly used (e.g., for deep learning). In the first two examples, while the article discusses them in the context of summation, we use FedAvg to keep the implementation as simple as possible, since the same operations can be performed using FedAvg as well. In the third example (SecureSum), we show how to implement a custom strategy specifically for federated summation.

The examples are implemented using the simulation mode of the Federated Learning framework Flower. For more information on Flower, we refer the reader to Flower's webpage.

Create the conda environment

To run the examples in this Project install all the dependencies in a conda environment by running the following commands.

conda env create -f FL_Bio.yaml

Next, activate the environment by running:

conda activate FL_Bio