RestingStateMultiband

Scripts supporting "Which multiband factor should you choose for your resting-state fMRI study?"

Manuscript, Web Supplement, and the table of gfactors are included here (see /ManuscriptSupplementGit/). Manuscript also available from NeuroImage: https://doi.org/10.1016/j.neuroimage.2021.117965

Data are available at https://openneuro.org/datasets/ds003540. Note scripts include here need to be adapted for the files on OpenNeuro, which were renamed using the BIDS standard.

A) Shell scripts. These must be run first:

1. preprocessing_loopSubjects.sh loops through the subjects and creates a copy of preprocessing_singlesubj_3dTproject.sh for each subject in which the "subjectID" is replaced with the numeric value. It then submits (via qsub) the preprocessing scripts for all subjects. The script performs preprocessing for four pipelines: without / with temporal filtering x without / with spatial smoothing. The main preprocessing script utlizes elements of the Functional Connectomes 1000 project scripts (https://www.nitrc.org/projects/fcon_1000/) but with a number of updates including topup distortion correction, fnirt, and simultaneous filtering and nuisance regression with AFNI 3dTproject. This takes approximately four hours (assuming the number of nodes=number of subjects).

2. Voxel2nodesystem_loopSubjects.sh calculates the average time course for each power ROI. Loops across subjects.

   Creates files Voxel2nodesystem_singlesubj.sh and Voxel2nodesystem_singlesubj.m where the former is used to qsub the latter. Also calculates the subject correlation matrices. Takes about 15 minutes.

B) R and Matlab Scripts: these were run interactively

1_MotionExamine_cluster.R: flags scans that failed motion control

2_CreateAveSDMaps.R: creates standard deviation maps

3_GFactorTables.R create g-factor tables

4_run_SeedMaps.m calls the function SeedMaps_create.m, creating seed maps for the specified ROIs and each pipeline

5_ConsolidateCorrMat_cluster.m gathers the subject correlation matrices into a single array. Also calcualtes the g*-factors.

6a_CorrelationsVGfactorsAndMB_GAMsGEEs.R Analyzes impacts of g-factor and multiband factor for a subset of intra-commmunity edges (6 regions/communities)

1. GAMMs analyzing impact of gfactor on correlations
2. GEEs analyzing subset of edges for different communities and MB factor Creates figures "Z correlation versus g*-factor" from gamm

6b_CorrelationsVGfactorsAndMB_GAMsGEEs_ss.R same as 6a but run on the datasets with 6-mm smoothing

7a_ExamineCorrelationMatrices_plots.R -creates plots of MB factors versus SB 2 mm colored by g* factor -creates plots of correlation matrices -creates the dataset somemb.z.cor.srt.RData for input to the permutation tests

7b_ExamineCorrelationMatrices_plots_sm.R same as 5a but run on the datasets with 6-mm smoothing

C) Permutation tests were run on the cluster using qsub, e.g.,

cd ~/risk_share/OptimalSMS_rsfMRI/Programs
qsub -cwd -N permtest_notf a6a_CompareMBFactor_PermutationTets_notf.sh

a8a_CompareMBFactors_PermutationTests_notf.sh simple shell script to execute using the scheduler and qsub

8a_CompareMBFactors_PermutationTests_notf.R permutation tests comparing proportion activated in MB 8 to other acquisitions Takes approximately 2.25 hours to run

a8a_CompareMBFactors_PermutationTests_notf.sh

8b_CompareMBFactors_PermutationTests_tf.R

a8c_CompareMBFactors_PermutationTests_notf_vs_tf

8c_CompareMBFactors_PermutationTests_notf_versus_tf Permutation tests to support the hypothesis that temporal filtering decreases the proportion active

a8d_CompareMBFactors_PermutationTests_notf_sm.sh simple shell script to execute using the scheduler and qsub

8d_CompareMBFactors_PermutationTests_notf_sm.R identical to 6a but using smoothed data

D) This script is run interactively:

9_PermutationTests_tables.R creates tables summarizing the permutation tests from 8a-8d.