Federated Biobank Project

A copy of repository https://github.com/genxnetwork/fl-genomics, which explores federated learning for phenotype-from-genotype and ancestry-from-genotype prediction. See the paper at https://www.medrxiv.org/content/10.1101/2023.01.24.23284898v2

TG Environment Installation

It's recommended to work with TG codebase using conda environemnt.

1. Install conda: https://docs.conda.io/projects/conda/en/latest/user-guide/install/linux.html.

2. Activate conda environment suing requirenemnts file requirements_tg.txt:

conda create --name genx --file tg_requirements.txt

3. Activate the environment:

conda activate genx

4. Install pgenlib from PLINK's repo:

git clone https://github.com/chrchang/plink-ng.git
cd plink-ng/2.0/Python
python3 setup.py build_ext
pip install -e .

Structure

• split module generates node datasets from the whole UKB dataset based on self-reported ancestry.
• qc module encapsulates node-based quality control.
• dimred module performs different strategies of dimensionality reduction.
• fl module compares various FL strategies on selected SNPs.

Visualisation

Dash App

Run dash_app.py and open the link that appears in console in a browser. There assign filter+value or graph elements (x-axis, y-axis, color, etc.) to columns via dropdowns. Then press submit.