lilythepooh / Scalable_Bigraphical_Lasso

Scalable Bigraphical Lasso code

This repository reproduces the research presented in the paper: "Two-way Sparse Network Inference for Count Data", Authors: Li, Sijia; López-García, Martín; Lawrence, Neil D; Cutillo, Luisa, accepted at AISTATS2022.

Matlab libraries needed:

We are using Matlab 2020b. You will need to install the following matlab libraries: ndlutil, rca, glmnet_matlab, L1General. Copies of these are included in this repo. You will also need the library of Teralasso (Greenewald et al. 2019) for some comparison studies.

We assume that you'll run these scripts from your current working directory. Change the mypath variable if you do otherwise.

The experiment was run on Windows 10. Please let us know if you encounter any problems either on Windows or other operating system.

For Figure 1

• run Figure1.m.

For Figure 2

• run run_syntheticGaussian_Three_method_time.m.

For Figure 3, Figure 9 and Figure 10

• run run_synthetic_data.m.

For Figure 4

• run run_synthetic_data_three_blocks.m.

For Figure 5

• run main_genes.m.

For Figure 6

• run main.R.