Analysis code for: "Dynamic network features of functional and structural brain networks support visual working memory in aging adults"

This repository contains the code used for analyses in:

Neudorf, J., Shen, K., & McIntosh, A. R. (2024). Dynamic network features of functional and structural brain networks support visual working memory in aging adults. bioRxiv. https://doi.org/10.1101/2024.07.30.605891

Environment installation instructions:

Python:

Local Bash

git clone https://github.com/McIntosh-Lab/dFC_Cam-CAN.git
cd dFC_Cam-CAN
conda create -n neudorf_dFC -c conda-forge python=3.10 numpy scipy nibabel nilearn matplotlib pillow pandas seaborn tqdm statsmodels plotnine
conda activate neudorf_dFC
pip install nctpy
python -m pip install matlabengine=9.13.11 #matlab 2022b.2
cd python_packages/brainvistools
python -m pip install .
cd ../PLS_wrapper
python -m pip install .
cd ../PyNeudorf
python -m pip install .

HPC (Alliance Canada) Bash

git clone https://github.com/McIntosh-Lab/dFC_Cam-CAN.git
cd dFC_Cam-CAN
module load StdEnv/2020 matlab/2022b.2 python/3.10
python3.10 -m venv neudorf_dFC
source neudorf_dFC/bin/activate
pip install numpy scipy nibabel nilearn matplotlib==3.6.2 pillow pandas seaborn==0.12.1 tqdm statsmodels plotnine==0.12.3 certifi
pip install nctpy
#edit line below by finding matlabroot using `which matlab`, and substitute `matlab` executable with `glnxa64`
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/cvmfs/restricted.computecanada.ca/easybuild/software/2020/x86-64-v3/Core/matlab/2022b.2/bin/glnxa64
#set matlabengine version number to match matlab version 2022b.2
python -m pip install matlabengine==9.13.11
cd python_packages/brainvistools
python -m pip install .
cd ../PLS_wrapper
python -m pip install .
cd ../PyNeudorf
python -m pip install . 

R:

Local Bash

Rscript install_packages.r

HPC (Alliance Canada) Bash

module load StdEnv/2020 r/4.3.1 gcc/9.3.0 gdal/3.5.1 udunits/2.2.28
mkdir -p ~/.local/R/$EBVERSIONR
export R_LIBS=~/.local/R/$EBVERSIONR/
Rscript install_packages.r

Matlab:

Local Bash

Follow instructions here and add plscmd folder to startup.m: https://github.com/McIntosh-Lab/PLS

HPC (Alliance Canada) Bash

These commands should be run in the main dFC_Cam-CAN directory.

git clone https://github.com/McIntosh-Lab/PLS
cp -r PLS/plscmd ~/matlab
echo "addpath(genpath('~/matlab/plscmd'))" >> ~/matlab/startup.m
echo "addpath(genpath('${PWD}/python_packages/PLS_wrapper'))" >> ~/matlab/startup.m

Usage

LEiDA results used in paper are in matlab_toolboxes/leida-matlab-1.0/res_Cam-CAN_TVB_SchaeferTian_218

PLS in outputs/leida-matlab/PLS, PLS_group_analyses, and outputs/nctpy/5/PLS

NCT outputs in outputs/nctpy

For code review: Get data from ~/projects/def-rmcintos/jneudorf/Cam-CAN/code_review (behavioural data, SC matrices and rsfMRI timecourses) Extract data/ folder from dFC_data.tar.gz into main folder. To reproduce:

1. Figure 4 PLS behavioural correlation (A; outputs/leida-matlab/PLS/TP_dict_K5_age_1000_its_197_subs_0_to_150_age_range/TP_dict_K5_age_1000_its_197_subs_0_to_150_age_range_lv1_behav_corr.png) and BSR transition probability matrix (B; outputs/leida-matlab/PLS/TP_dict_K5_age_1000_its_197_subs_0_to_150_age_range/TP_dict_K5_age_1000_its_197_subs_0_to_150_age_range_lv1_bsr_matrix.png)
2. Figure 5 young adult behavioural correlation (B; outputs/nctpy/5/PLS/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_114_subs_0_to_50_age_range/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_114_subs_0_to_50_age_range_lv1_behav_corr.png) and brain map BSR (C; outputs/nctpy/5/PLS/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_114_subs_0_to_50_age_range/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_114_subs_0_to_50_age_range_lv1_bsr_cortex.png and outputs/nctpy/5/PLS/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_114_subs_0_to_50_age_range/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_114_subs_0_to_50_age_range_lv1_bsr_subcortex.png) plus older adult behavioural correlation (D; outputs/nctpy/5/PLS/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_83_subs_50_to_150_age_range/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_83_subs_50_to_150_age_range_lv1_behav_corr.png) and brain map BSR (E; outputs/nctpy/5/PLS/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_83_subs_50_to_150_age_range/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_83_subs_50_to_150_age_range_lv1_bsr_cortex.png and outputs/nctpy/5/PLS/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_83_subs_50_to_150_age_range/state_transition_5_to_3_node_energy_dict_continous_age_Prcsn_PerceptionTest_1000_its_83_subs_50_to_150_age_range_lv1_bsr_subcortex.png) Note that the exact BSRs will be different than in the paper because of resampling. PLS outputs saved in outputs/leida-matlab/PLS and outputs/nctpy/5/PLS

Order to run code:

1. consistency_thresholding.py
2. import_Cam-CAN_data.py
3. prep_data.py
4. LEiDA toolbox in matlab_toolboxes/leida-matlab-1.0/. Info is in README.md as well, but edit paths at the top of LEiDA_Start.m and run_all_after_start.m. Then run LEiDA_Start.m followed by run_all_after_start.m. You can change n_permutations and n_bootstraps to a lower number for testing to make these run faster.
5. leida-matlab_analyses.py will perform PLS analyses on the LEiDA outputs and produce images used in figures 2, 3, and 4.
6. nctpy_analyses.py will perform the network control theory simulation and the PLS analysis of the state 5 to 3 transition (figure 5).
7. PLS_group_analyses/PLS_group_analyses.m performs secondary mean-centred PLS analyses for LEiDA and NCT analyses.

For HPC (Alliance Canada), use sbatch submission scripts provided. These scripts should be submitted as a job while in the main dFC_Cam-CAN directory, unless otherwise specified. In order to submit these scripts as jobs, please use the sbatch command (see first bullet for an example):

• sbatch 1_2_3_submit_preprocessing.sh
• change directories with cd matlab_toolboxes/leida-matlab-1.0 then sbatch 4_submit_leida.sh (refer to step 4. above and edit paths. This script should be submitted as a job while in the matlab_toolboxes/leida-matlab-1.0/ directory.) When finished change directories back to project directory with cd ../...
• sbatch 5_submit_leida_analyses.sh
• sbatch 6_submit_nctpy_analyses.sh (change RSCRIPT='/usr/bin/Rscript' at top of nctpy_analyses.py to RSCRIPT='/cvmfs/soft.computecanada.ca/easybuild/software/2020/avx2/Core/r/4.3.1/bin/Rscript')
• change directories with cd PLS_group_analyses then sbatch 7_submit_group_analysis.sh (This script should be submitted as a job while in the PLS_group_analyses/ directory.)