Repositiry for the article: "Gene regulatory network inference using mixed-norms regularized multivariate model with covariance selection" by Alain Mbebi & Zoran Nikoloski

This is the repository for the manuscript "Gene regulatory network inference using mixed-norms regularized multivariate model with covariance selection" by Alain J. Mbebi & Zoran Nikoloski.

Organisation

The folder Codes contains the following R scripts with the K-folds cross-validation option to learn the hyperparameters:

Mixed_L1L21_GRN.R which computes L1L21-solution

Mixed_L1L21G_GRN.R which computes L1L21G-solution

Mixed_L2L21_GRN.R which computes L2L21-solution

Mixed_L2L21G_GRN.R which computes L2L21G-solution

L1L21_Dream5_Scerevisiae_example_run.R is an example run using the L1L21-solution with S. cerevisiae data (Network 4 in DREAM5 challenge) All files needed to successfully run "L1L21_Dream5_Scerevisiae_example_run" are locaded in the folder Codes.

2. The folder Figures contains all figures in the manuscript.

3. The folder Inferred-networks contains all network objects for each dataset and each inference methods in the comparative analysis.

Dependencies and required packages

The following packages are required for the contending approaches in the comparative analysis: "devtools", "foreach", "plyr", "glmnet" and "randomForest".

GENIE3

The GENIE3 package can be installed from: http://bioconductor.org/packages/release/bioc/html/GENIE3.html

TIGRESS

The TIGRESS repository can be obtained from: https://github.com/jpvert/tigress

ENNET

The ENNET repository can be obtained from: https://github.com/slawekj/ennet

PLSNET

The Matlab source code of PLSNET can be obtained from: https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-016-1398-6#Sec17

PORTIA

The PORTIA repository can be obtained from: https://github.com/AntoinePassemiers/PORTIA

D3GRN

The Matlab source code of D3GRN can be obtained from: https://github.com/chenxofhit/D3GRN

Fused-LASSO

The fused-LASSO repository can be obtained from: https://github.com/omranian/inference-of-GRN-using-Fused-LASSO

ANOVerence

Because of some technical issues (e.g code's accessibility: http://www2.bio.ifi.lmu.de/˜kueffner/anova.tar.gz), we were not able to reproduce ANOVerence results and used the inferred network from DREAM5 challenge instead.

4. Although the codes here were tested on Fedora 29 (Workstation Edition) using R (version 4.2.2), they can run under any Linux or Windows OS distributions, as long as all the required packages are compatible with the desired R version.