Split Knockoffs

This is a Matlab package to reproduce the experiments in the paper:

Yang Cao, Xinwei Sun and Yuan Yao, Controlling the False Discovery Rate in Structural Sparsity: Split Knockoffs, arXiv:2103.16159.

Dependency

• Glmnet in Matlab

Installation Details

This package is tested in Matlab R2020a and R2020b on Windows 10 with the updated version of Glmnet. A fortran compiler is required for using Glmnet in Matlab. We tested the Intel Fortran Compiler (Version 2021.2.0) with Visual Studio 2019.

Usage

To finish the installation, please run 'startup.m'. For usage of this package, type 'help split_knockoffs.filter' and 'help split_knockoffs.cv_filter' in the command line of Matlab.

Acknowledgement

This package adapts the '+knockoffs' folder from Knockoffs for matlab for common functions and comparisons.

Reproducible Experiments

• To reproduce the results of simulation experiments in our paper, please check the respective scripts in the 'simu_experiments' folder for details.
• To reproduce the results of experiments on Alzheimer‘s Disease in our paper, please check the respective scripts in the 'AD_experiments' folder for details.