This repository contains the R code for the simulation study for comparison of several methods for estimation of differential Gaussian graphical models.

Generate data folder has the code to generate all pairs of $G^{(1)}$ and $G^{(2)}$ graphs and corresponding precision and covariance matrices $\mathbf{\Omega}^{(1)}$, $\mathbf{\Omega}^{(2)}$, $\mathbf{\Sigma}^{(1)}$, $\mathbf{\Sigma}^{(2)}$ that we used in our simulation. Obtained graphs and matrices are also provided. Replications of the datasets $\mathbf{X}^{(1)}$, $\mathbf{X}^{(2)}$ are too heavy to upload but the code for their generation is provided and results do not vary a lot depending on the realization.

Methods folder contains the code we used to produce power-FDR curves for all methods and the performance results we obtained for each run for each method.

Plotting folder has the dataset with all the performance results combined and the code that we used for plotting.

Real data folder contains the dataset we used for the real data example and the code to produce the illustration.