A new Similarity Based Adapted Louvain Algorithm (SIMBA) for active module identification in p-value attributed biological networks

This repository contains the code needed to execute the SIMBA algorithm and reproduice the results presented in Singlan, N., Abou Choucha, F. & Pasquier, C. A new Similarity Based Adapted Louvain Algorithm (SIMBA) for active module identification in p-value attributed biological networks. Sci Rep 15, 11360 (2025). https://doi.org/10.1038/s41598-025-95749-6.

Table of contents

• Installation
• Reproduce paper results
• Usage on your own data
• Repository structure
• Data
• Requirements
• Cite
• Contact

Installation

To use this algorithm, it is needed to have Python 3.11 installed on your machine. Moreover, it is needed to install the required packages described in the Requirements section and listed in the environment.yml file.

To run the algorithm, it is needed to clone this repository, install the required packages and run the code using the commands described in the Reproduce paper results and the Usage on your own data sections.

Reproduce paper results

To reproduce the results of the paper on 'Fully simulated datasets you can use the following command:

./launch_simulated.sh

To reproduce the results of the paper on 'Rewire datasets' you can use the following command:

./launch_rewire.sh

To reproduce the results of the paper on 'Value simulated dataset' you can use the following command:

./launch_value.sh

To reproduce the results of the paper on 'Real dataset' you can use the following command:

./launch_real.sh

Usage on your own data

To run the code on your own data, you can use the following command:

python main.py -d ./data/your_data

Available options are:

• -d or --data: Required the path to the data file.
• -gt or --ground_truth: Optional flag to indicate that graph contains ground truth.
• -n or --name: Optional flag to indicate that graph contains node names.
• -res or --results: Optional the path to the results directory. Default is ./results/.
• -o or --output: Optional the path to the output file. Default is ./results/output.txt.
• -min or --min: Optional the minimum size of the communities. Default is 5.
• -no_filter pr --no_filter: Optional flag to indicate that the filter should not be applied.
• -c or --clustering: Optional the clustering algorithm to use. Default is similarity. Available options are similarity, louvain ang None.
• -p or --priority: Optional the priority to use in the clustering algorithm (only for similarity-based). Default is worst. Available options are worst and best.
• -cs or --community_selection: Optional flag to indicate that the community selection should be applied.
• -t or --threshold: Optional the threshold to use in the community selection. Default is 0.05.
• -k or --k: Optional the number of communities to find (used in metrics calculation). Default is None.
• -v or --verbose: Optional flag to indicate that the verbose mode should be activated.

Repository structure

The repository is structured as follows:

• data/: contains the data used in the experiments.
• clustering/: contains the code of the similarity-based clustering algorithm.
• graph/: contains the code to represent the graph structure and the community detection algorithms.
• utils/: contains the code of the utility functions used in the experiments.
• Real_Graphs_Result/: contains the results of the real datasets experiments.
• main.py: contains the main code to run the experiments.
• Supplementary_Table_S13_all_results.xlsx: contains the results of the paper.

Data

Available data

The data used in the experiments are available in the data/ folder. Here is the list of the available data:

• [One_Cluster_Dataset.zip](data/One_Cluster_Dataset.zip): contains 1000 graphs of 1000 nodes with 1 community of 10 nodes to find.
• [Two_Clusters_Dataset.zip](data/Two_Clusters_Dataset.zip): contains 1000 graphs of 1000 nodes with 2 communities of 10 nodes to find.
• [Three_Clusters_Dataset.zip](data/Three_Clusters_Dataset.zip): contains 1000 graphs of 1000 nodes with 3 communities of 10 nodes to find.
• [Ten_Clusters_Dataset.zip](data/Ten_Clusters_Dataset.zip): contains 1000 graphs of 1000 nodes with 10 communities of 10 nodes to find.
• [Medium_Dataset.zip](data/Medium_Dataset.zip): contains 100 graphs of 6344 nodes with 10 communities of 10 nodes to find.
• [Dataset_Rewire_on_Human_0-99_CutOff.zip](data/Dataset_Rewire_on_Human_0-99_CutOff.zip): contains 100 graphs of 6344 nodes with 10 communities of 10 nodes to find.
• [Dataset_Rewire_on_Human_0-8_CutOff.zip](data/Dataset_Rewire_on_Human_0-8_CutOff.zip): contains 100 graphs of 14219 nodes with 50 communities of 15 nodes to find.
• [Tammaro_2024_brain_STRING_v12.npz](data/Tammaro_2024_brain_STRING_v12.npz): contains the graph obtained from article: HDAC1/2 inhibitor therapy improves multiple organ systems in aged mice (Tammaro, 2024).
• [Tammaro_2024_heart_STRING_v12.npz](data/Tammaro_2024_heart_STRING_v12.npz): contains the graph obtained from article: HDAC1/2 inhibitor therapy improves multiple organ systems in aged mice (Tammaro, 2024).
• [Tammaro_2024_kidney_STRING_v12.npz](data/Tammaro_2024_kidney_STRING_v12.npz): contains the graph obtained from article: HDAC1/2 inhibitor therapy improves multiple organ systems in aged mice (Tammaro, 2024).

Data format

Each graph is stored in a .npz file. The file contains the following keys:

• adjacency_data: the adjacency sparse matrix of the graph.
• adjacency_indices: the indices of the adjacency sparse matrix of the graph.
• adjacency_indptr: the indptr of the adjacency sparse matrix of the graph.
• adjacency_shape: the shape of the adjacency sparse matrix of the graph.
• feature_data: the feature sparse matrix of the graph.
• feature_indices: the indices of the feature sparse matrix of the graph.
• feature_indptr: the indptr of the feature sparse matrix of the graph.
• feature_shape: the shape of the feature sparse matrix of the graph.
• labels: Optional - Use with the -gt option the ground truth of the graph.
• label_indices: Optional - Use with the -gt option the ground truth indices of the graph.
• name: Optional - Use with the -n option the name of the nodes.

Requirements

The code is written in Python 3.11.

The following packages are required to run the code:

• scipy
• scikit-learn
• numpy
• pyunionfind
• scikit-network

Cite

If you use this code, please cite the following paper:

@article{Singlan2025,
   author = {Nina Singlan and Fadi Abou Choucha and Claude Pasquier},
   doi = {10.1038/s41598-025-95749-6},
   issn = {2045-2322},
   issue = {1},
   journal = {Scientific Reports 2025 15:1},
   keywords = {Bioinformatics,Gene expression analysis,Similarity-based clustering},
   month = {4},
   pages = {1-16},
   publisher = {Nature Publishing Group},
   title = {A new Similarity Based Adapted Louvain Algorithm (SIMBA) for active module identification in p-value attributed biological networks},
   volume = {15},
   url = {https://www.nature.com/articles/s41598-025-95749-6},
   year = {2025}
}

Contact

If you have any question, please contact me at singlan.nina@gmail.com