Prioritization of master regulators through a Boolean network

(c) Clémence Réda, 2023.

Due to the presence of copyrighted databases, the license for this code is Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).

Due to the presence of copyrighted databases, the license for this code is Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).

Environment

Create two Conda environments using the two scripts install_env_application.sh (environment "application_envir") and install_env_synthesis.sh (environment "synthesis_envir")

Data (refractory epilepsy)

Import the initial states from Mirza et al., 2017 and the M30 genes from Delahaye-Duriez et al., 2016

conda activate synthesis_envir
python3 download_Refractory_Epilepsy_Data.py
conda deactivate

Building a Boolean network

You need to register to the LINCS L1000 database and the DisGeNet database and write up the corresponding credentials and API keys to files utils/credentials/credentials_LINCS.py and utils/credentials_DISGENET. Write up your name in utils/credentials/credentials_STRING.py.

Then check out the params.py file, where some parameters might need to be changed (documentation therein). Then run the following commands

conda activate synthesis_envir
python3 iter_main.py ## takes some time
python3 model_selection.py ## final model selection and network robustness plots
conda deactivate

The final solution is solution.bnet (with the initial values in params.py, this file is located at refractory_epilepsy/).

Detection of master regulators

Check out params.py to change parameter values, if needed.

conda activate application_envir
python3 application_regulators.py
conda deactivate

The result files are named application_regulators.csv, spread_values.csv and spread_values.rnk (with the initial values in params.py, these files are located at refractory_epilepsy/).

Network analysis with Cytoscape

Network analyses are performed with Cytoscape 3.8.0. You need to download the module CytoCtrlAnalyser (version 1.0.0). Then run

conda activate application_envir
python3 bonesis2cytoscape.py # converts graph files to Cytoscape-readable formats
conda deactivate

Compute at least MDS, ControlCentrality (in CytoCtrlAnalyzer) and Outdegree, Indegree (NetworkAnalyzer, built-in module in Cytoscape). You can run the following script to get source evidence annotations on edges from the STRING database

conda activate application_envir
python3 annotate_edges.py
conda deactivate

Visualization of gene measures & nrichment analyses

First run

conda activate application_envir
python3 compare_rankings.py
conda deactivate 

Then, use the online tool WebGestalt. Select the following parameters

• Organism of interest: Homo sapiens
• Method of interest: ORA
• Functional Database: disease > DisGeNet
• Select Gene ID Type: Gene symbol
• Upload Gene List: upload the list at refractory_epilepsy/rankings/values/hipp_epileptic_list.txt
• Upload User Reference Set: Gene symbol
• File and Select ID type: refractory_epilepsy/rankings/values/hipp_epileptic_bkgrnd.txt

In "Advanced parameters", leave default parameters, except for

• Significance Level: FDR > 0.2

Download the corresponding .zip file and put it in refractory_epilepsy/rankings/enrichments/ by using the same type of filename syntax as shown in the examples (note: if the enrichment files cannot be found, have a look at compare_rankings.py).

Running on the set of 50 solutions

Copy the solutions in folder refractory_epilepsy/25additionalsolutions/ in refractory_epilepsy/, and run (NOTE THAT THIS ERASES FOLDER PLOTS/ AND FILES SOLUTION.BNET, COLLAPSED_MODEL.CSV)

conda activate synthesis_envir
python3 model_selection.py
conda deactivate

Pull requests, issues, suggestions?

[email protected]