Multilayer AD

This repository contains the code related to the analysis and simulation of networks for our research paper.

Important note on data access: The raw data used in this study were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. To access the same dataset, researchers must apply for access through the ADNI data portal: https://adni.loni.usc.edu/

Once you are granted access, you will be able to download the necessary files and reproduce the results using the code provided in this repository.

Description of the datasets

• description_datasets.ipynb: Creates the tables in the SI with all the nodes, their description, and the number in the nodes labels.

Node Dictionary Creation

• node_dict.py: Creates the node_dictionary.pickle file, where the keys are the layers and the values are the nodes.

Mutual Information

Individual Networks

• mi_ind.py: Computes the MI of cross networks and performs the permutation test.
• mi_ind_net.py: Builds the individual networks from the files generated by mi_ind.py.

Cross Networks

• mi_cross.py: Cmputes the MI of cross networks and performs the permutation test.
• mi_cross_net.py: Builds the cross networks from the files generated by mi_cross.py.

Individual Networks for nl, mci, and dementia

• mi_ind_diag.py: Computes the MI of individual networks for nl, mci, and dementia, and performs the permutation test.
• mi_ind_diag_net.py: Builds the individual networks for nl, mci, and dementia from the files generated by mi_ind_diag.py.

Cross Networks for nl, mci, and dementia

• mi_cross_diag.py: Computes the MI of cross networks for nl, mci, and dementia, and performs the permutation test.
• mi_cross_diag_net.py: Builds the cross networks for nl, mci, and dementia from the files generated by mi_cross_diag.py.

Global and Multilayer Network Analysis

• bignet_analysis.ipynb: Analyzes individual networks and creates the global and multilayer networks.

Calculation of the 10 Shortest Paths

• make_paths.py: Computes the 10 shortest paths for each source and target node in the global or multilayer network.
• make_paths_diag.py: Computes the 10 shortest paths for each source and target node in the nl, mci, and dementia networks.

Shortest Paths Analysis and Plots

• paths_analysis.ipynb: Analyzes the shortest paths and generates plots.

Connectivity and Density Matrices Analysis

• cross_analysis.ipynb: Computes the density (connectivity) matrices.

FDG PET connections with nodes from other layers

• table_PET.ipynb: Creates the table with the FDG PET connections with nodes from other layers and their MI.

Pearson

• pearson_ind.py: Computes the Pearson correlation coefficient for individual networks.
• pearson_ind_net.py: Builds the individual networks from the files generated by pearson_ind.py.
• pearson_cross.py: Computes the Pearson correlation coefficient for cross networks.
• pearson_cross_net.py: Builds the cross networks from the files generated by pearson_cross.py.
• pearson_ind_diag.py: Computes the Pearson correlation coefficient for the individual networks of nl, mci, and dementia.
• pearson_ind_net_diag.py: Builds the individual networks of nl, mci, and dementia from the files generated by pearson_ind_diag.py.
• pearson_cross_diag.py: Computes the Pearson correlation coefficient for the cross networks of nl, mci, and dementia.
• pearson_cross_net_diag.py: Builds the cross networks of nl, mci, and dementia from the files generated by pearson_cross_diag.py.
• bignet_analysis_pearson.ipynb: Analyzes the individual networks and creates the global and multilayer networks, including the global networks for nl, mci, and dementia.

Boolean Simulations

• boolean_simulations.py: Performs Boolean simulations with an option to adjust the noise level.
• boolean_simulations_diag.py: Performs Boolean simulations for nl, mci, and dementia with an option to adjust the noise level.
• cross_corr.py: Computes the cross-correlation coefficient (maximum value of the temporal cross-correlation across all lag times).
• cross_corr_diag.py: Computes the cross-correlation coefficient for nl, mci, and dementia.
• noise_analysis.ipynb: Analyzes the effect of noise on the cross-correlation coefficient.
• paths_cross_corr.py: Builds the network from the cross-correlation coefficient and computes the best shortest paths.
• paths_cross_corr_diag.py: Builds the network from the cross-correlation coefficient and computes the best shortest paths for nl, mci, and dementia.
• permut_bool.py: Performs negative control by permuting network connections and computing Boolean simulations and best paths for each permutation.
• analysis_paths.ipynb: Analyzes the paths obtained in the dynamic analysis. For negative controls: counts the frequency of each path after permutations.
• create_network_paths.ipynb: From the CSV file with paths that passed the negative control, builds the networks to be visualized in Cytoscape.

Installation

Requirements

• Python 3.14+
• pip (latest version recommended)

Steps

1. Clone this repository
2. Create a virtual environment (optional but recommended)

python -m venv .venv

PowerShell:

.\.venv\Scripts\Activate.ps1

Bash:

source .venv/bin/activate

3. Install the dependencies

python -m pip install --upgrade pip
pip install -r requirements.txt