Group-specific discriminant analysis

Introduction

This repository contains the implementation of Group-Specific Discriminant Analysis (GSDA) and experiments from the GigaScience paper Group-specific discriminant analysis enhances detection of sex differences in brain functional network lateralization by Zhou et al. (2025).

Framework

Datasets

The resting-state fMRI data from HCP [1] and GSP [2] is used in this study. Code for data preprocessing is available at /preprocess. Processed data is available at Zenodo: [HCP], [GSP].

System Requirements

numpy>=1.24.3
pandas>=1.5.3
scipy>=1.10.1
scikit-learn>=1.2.2
pytorch>=2.0.0
yacs

Installation Guide

pip install -r requirements.txt

Instructions for Use

Basic usage:

python main.py --cfg configs/demo-hcp.yaml

Please create more .yaml files for different random seeds and datasets.

Demo

We provide GSDA running demo through a cloud Jupyter notebook on . Note the number of repetition is limited for faster demonstrations. This demo takes 10-20 minutes to complete the training and testing process.

References

[1] Smith, S. M. et al. Resting-state fMRI in the human connectome project. NeuroImage 80, 144–168 (2013)

[2] Holmes, A. J. et al. Brain genomics superstruct project initial data release with structural, functional, and behavioral measures. Sci. Data 2, 1–16 (2015)

Citation

If you use this code in your research, please cite the following paper:

@article{10.1093/gigascience/giaf082,
    author = {Zhou, Shuo and Luo, Junhao and Jiang, Yaya and Wang, Haolin and Lu, Haiping and Gong, Gaolang},
    title = {Group-specific discriminant analysis enhances detection of sex differences in brain functional network lateralization},
    journal = {GigaScience},
    volume = {14},
    pages = {giaf082},
    year = {2025},
    month = {08},
    issn = {2047-217X},
    doi = {10.1093/gigascience/giaf082},
    publisher={Oxford University Press}
}