%% READ ME
% Scripts to perform functional connectivity analysis. The scripts were % used in the paper: "Action experience in infancy predicts visual-motor % functional connectivity during action anticipation"
% IMPORTANT NOTE: These scripts use Acropolis, a server from the University % of Chicago. However, the scripts can be easily adapted to analyze the % data outside of Acropolis.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% To run jobs in parallel in ACROPOLIS - for Step 1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Go to the folder in which you have a submitloop (.sh) file % Open a terminal there and run the loop: ./nameloop.sh
% The loop will find all the files of a format you have specified (e.g. *.set) % in a folder you have specified. For each file, it will run the .qsub file of interest % (that is, several jobs will be sent to the cluster in parallel, one for each input file).
% The .qsub file will save in the environment a variable called % MATLABSUBJECT that will have the name of the analyzed file (with its % path). Then .qsub will indicate to run a specific script in matlab. This % script will read the name of the current file of analysis via sub = getenv('MATLABSUBJECT');
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% Things to change on the .sh and .qsub scripts %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % .sh: path folder with input files, file format of interest & name .qsub script % .qsub: # of processors and RAM memory to use + name matlab script to run
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% General structure for connectivity analysis %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% The process is organized in 4 steps (see below). % Input data: preprocessed artifact-free EEG data transformed to time-frequency
% Analysis: % 1. Whole-Brain: wholebrain_github.m (step 1) -- calculate adjacency matrices for each participant, % wholebrain_allsubjects_clusters_github.m (step 2) -- organize adjacency matrices of all participants together, % export_wholebrain_ISPCtime_github.m (step 3) -- export whole-brain data on organized tables to be analyzed in R. % Connectivity_analysis_LAEEG_FINAL_rev.Rmd (step 4) -- plot whole-brain data and do statistical analysis
% 2. Inter-site network (FO, FC, PO, PC, OC): % main_analysis_chanpair_github.m (step 1) -- calculate connectivity between two areas of interest % main_analysis_chanpair_acrosstime_github.m -- organize data of all participants together (step 2), % and export data (step 3) to be analyzed in R. % Connectivity_analysis_LAEEG_FINAL_rev.Rmd (step 4) -- plot inter-site connectivity data (normalized by whole-brain) and do statistical analysis
%%%%%%%%%% %% STEP 1. Connectivity analysis %%%%%%%%%% %% Channel Pair (connectivity between specific pairs): run main_analysis_chanpair_github.m % Don't need to run it through the Terminal (in parallel). You can just run % the matlab code, which will analyze the subjects sequently. % Save data at LAEEG_age/chanpair/ % Calculates connectivity either over trials (type = 1) or over time (type = 2) % between the clusters of interest (O-C, F-C, P-C, F-O, P-O) using ISPC % Dimensions conn over time: frequency, time, trial, condition % IMPORTANT: time frequency has been transformated, so it's % not the same as the standard saved for conn over trial. % Dimensions conn over trial: frequency, time, condition
%% Whole Brain connectivity:
% Here you run the jobs (wholebrain_github.m) in Acropolis (in parallel) through submitloop_wb
% Save data at LAEEG_age/wb_output/
% Calculates adjecency matrices (ISPC-time) between all electrode
% pairs + the threshold of whole brain connectivity (1 median + 1 SD).
% For each subj, you obtain 2 files: 2 conditions
% Outputs:
% ispc_connectivity: adjacency matrix (time x chan x chan)
% thres_... is the threshold value for each time point and
% New time vector
% IMPORTANT: You need to specify manually the frequency range of
% interest (theta, alpha, beta), and the age group to analyze (9m or 12m)
% In order to have all the data, you will have to run the script 6
% times: 3 for 9-month-olds (3 freqs) and 3 for 12-month-olds (3 freqs)
%%%%%%%%%% %% STEP 2. OPEN DATA OF ALL SUBJECTS & CONDITIONS AND SAVE IT TOGETHER %%%%%%%%%% % No need to use the terminal (run directly in matlab)
%% For whole brain connectivity:
% wholebrain_allsubjects_clusters_github.m (saves the threshold of
% connectivity calculated without clustering (NOcl) and also with
% clustering (cl), in which case the connectivity is first averaged
% across all the electrodes in the cluster and then the median + 1
% SD is calculated based on data of the 15 resulting clusters.
% You can also decide to calculate the average rather than the
% median. To do that, define thres_formula as 'avg_' rather than ''.
% The analysis will be performed for theta, alpha and beta.
% Ouput: LAEEG_age/wholebrain_all_acrosstime_age_condition
% Outcome1: adjacency matrices for each participant and time point
% When clustering, dimensions are: subject, time, cluster, cluster
% When not clustering: subject, time, channel, channel
% Outcome2: threshold (data we will used to normalize the connectivity
% of the channel pair and for whole brain analysis)
% Dimensions: subject x time x condition
% (condition 1 = CANE (novel); condition 2 = GRASP (familiar))
% IMPORTANT: Change the age of the participants depending on which
% group you want to analyze. Perform analysis both for 9m and 12m
%% For channel pair connectivity:
% organizes connectivity between channel pairs of all subjects. Also, it
% creates a table that will be used to analyze data in R.
% The analysis will be performed for theta, alpha and beta.
% Main scripts:
% main_analysis_chanpair_acrosstime_github.m
% Ouputs:
% Connectivity matrix for each channel pair: subject x time x
% condition (e.g. ispc_OC_all.mat)
% Connectivity matrix for each channel pair with trial
% information: subject x time x trial x condition (e.g. ispc_OC_trial.mat)
% List of subjects included
% Number of subjects who provided 3 trials, 4 trials and 5
% trials (e.g. sum_4trials.mat)
% It also creates a .csv for each channel pair with trial or
% without trial information that will be used to analyze data in R
%%%%%%%%%% %% STEP 3. Save whole brain data (threshold) for analysis in R %%%%%%%%%%
% Script: export_wholebrain_ISPCtime_github.m % Output: .csv file with the data of the 12mo and 9mo (threshold whole-brain)
%%%%%%%%%% %% STEP 4. Open data in R to do plots and statistical analysis %%%%%%%%%%
% Define the frequency band of interest and run the analysis. % In some cases, you may need to change some parameters, such as in which % time window you want to perform the analysis (e.g. Use window from -1000 % to -500 ms to test the relation between ISPC during anticipation and % grasp latency. % The code also includes some additional analysis not included in the % manuscript.