EC_network_tools

Tools for effective connectivity network generation and analysis used in the following publication:

Time-evolving controllability of effective connectivity networks during seizure progression. Brittany H. Scheid, Arian Ashourvan, Jennifer Stiso, Kathryn A. Davis, Fadi Mikhail, Fabio Pasqualetti, Brian Litt, Danielle S. Bassett. Proceedings of the National Academy of Sciences, Feb 2021, 118 (5) e2006436118; DOI: https://doi.org/10.1073/pnas.2006436118

Requirements

• Matlab >= 2018a
• R >= 3.0.0
• R libraries: bootnet, R.matlab
• The commandline tool "Rscript" must be installed and should be on the system PATH
• GenLouvain Version 2.2: https://github.com/GenLouvain/GenLouvain

Note, code in this repository has only been tested on linux and mac systems and may not be compatible with other environments.

Example

% Example EC Controllability Pipeline

addpath('helper_functions')

% load example grid recording data with artifact and strip channels removed
raw = load('HUP64_sz1_ictal.mat', 'd');

% process data
seizure = cleanRawSeizure(raw.d.data, raw.d.Fs);

% Get regularized partial correlation matrices across windowed data
timeseries_data = seizure; %(64 channels x 59255 samples)
windowSize = 500;
Networks = getNets('./', timeseries_data, windowSize, 0.5, [0, 0.1, .5]);

% Partition Network similarity matrix into desired number of communities
init_gamma = (0.8:0.05:1.2); % Set of initial spatial resolution parameters 
nTarget = 3;               % Target number of communities 
Qiter = 100;               % iterations of Louvain algorithm
Partitions = findSimComms(Networks.wSim_0_5, init_gamma, Qiter, nTarget);