Gene Network

Author: Alice Hermann

A short description: this project extends a gene list using StringDB and FunCoup. Optuna, an optimization framework, is used to automatically tune parameters for network expansion, filtering, and robustness assessment. The optimization is based on F1-score. It requires a configuration file (JSON) specifying:

• the input file with a gene list (one gene per line)
• the number of trials and the number of runs per trial for Optuna
• the name of the output file (XLSX)
• the name of the output graph (PNG)

The added genes in the network are saved in a file with information from UniProt.

---

Running the project

Run the main script with your configuration file:

python main.py config.json

---

Example of configuration file

{
"input_file": "datas/genes.txt",
"trials_number": 100,
"runs_number": 10,
"output_file": "results/output.xlsx",
"output_graph": "results/network.png"
}

---

Example of output file

In the excel file:

• gene name
• degree centrality for the gene
• betweenness centrality for the gene
• closeness centrality
• eigenvector centrality
• degree of the gene
• number of direct neighbours
• interaction type
• number of direct connections with genes from original list
• neighbours from original list
• source number for this interaction
• sources for these interactions
• information via UniProt (activity regulation, pathway, structure, tissue specificity, ...)

The PNG is the visualization of the gene network:

• Red nodes: original genes from the gene list
• White nodes: added genes

---

Requirements

This project requires the following Python packages:

• Optuna
• requests
• NetworkX
• NumPy
• pandas
• scikit-learn

Install via:

pip install -r requirements.txt

---

Project Structure

• src/ : source code
• datas/ : gene lists
• results/ : results files
• config/ : configuration files

---

Notes

• The configuration file is mandatory for running main.py.
• The output includes both an XLSX file and a PNG graph of the network.

EOF