Introduction

These scripts were used to do some preliminary analyses in the multi flip angle data shared by University of Allahabad and previously described on the following pre-print from that group: https://www.biorxiv.org/content/10.1101/2020.07.09.196568v1

The scripts are created with the following in mind:

1. PRJDIR contains the project directory

2. Within the project directory data is expected to reside in folder PrcsData

3. Inside PrcsData there is one directory per subject

4. Original EPI and T1 Data are expected to be in D00_OriginalData

5. Files inside D00_OriginalData are expected to have the following names:

   • P001_Anat+orig.BRIK/HEAD: This is the T1 high resolution anatomical for subject ID P001
   • P001_fMRI_015+orig.BRIK/HEAD: This is the rest scan for FA = 15 for subject ID P001
   • P001_fMRI_050+orig.BRIK/HEAD: This is the rest scan for FA = 50 for subject ID P001
   • P001_fMRI_077+orig.BRIK/HEAD: This is the rest scan for FA = 77 for subject ID P001
   • P001_fMRI_090+orig.BRIK/HEAD: This is the rest scan for FA = 90 for subject ID P001

6. Preprocessing of T1 datasets will occur in D01_Anatomical

7. Preprocessing of EPI data will occur in a different directory per flip angle : D03_Preproc_NoRICOR_FIR_${FA}

8. There are two type of scripts: bash scripts and jupyter notebooks

How to run the scripts

Scripts are expected to be run in order, e.g., FA02_XXX, FA03_XX, etc.

Here is a basic description of each script set.

1. FA02_Preproc_Anat: This script set will pre-process the anatomical scans. This includes skull striping, bias correction, non-linear transformation to MNI space and tissue segmentation.

2. FA03_PrepareRefVols4Alignment: This script set will generate an epi reference volume per EPI run. This refernece volume will be a bias corrected version of the first volume of each run.

3. FA04_Preproc_fMRI_NoRICOR_PerRun_FIR: This script calls afni_proc to generate pre-processing scripts for each run separately. We will do analyses in original EPI space to minimize partial voluming. This script will do (among other things), the following: time shift correction, motion correction, compute registration to T1, spatial smoothing, create of masks for different tissues, and nuisance regression.

4. FA05_CreateNiis_and_Masks: This script set does a few extra operations following afni_proc. Those include: transformation of some HEAD/BRIK files into nifti (so that the can be loaded in Python later), aligment of 100 and 200 ROI Shaefer atlas to each fully pre-processed EPI dataset, alignment of 7 and 17 network yeo atlas to each fully pre-processed EPI dataset.

5. FA06a_ComputeTSNR_perTissue_OrigMethod: This script set computes TSNR in GM, WM, CSF (and eroded versions of these masks). It computes the TSNR after motion correction, spatial smoothing and nuisance regression.

6. FA07a_Compute_3dTCorrMaps: This script set computes maps of voxel-wise connectivity using AFNI program 3dTcorrmap.

7. FA07b_Compute_3dTCorrMaps_ShowResults_AvgMaps: This script set will bring the maps obtained in 6) to MNI space, so that we can generate an average connectivity map across all subjects in 8).

8. FA07c_Compute_3dTCorrMaps_AvgAcrossSubject.sh: This script averages all the subject individual voxel-wise connectivity maps

9. FA07d_Voxelwise_Connectivity_Results_GM: This notebook shows summary figures with changes in voxel-wise connectivity across the whole GM ribbon. It also include some basic preliminary statistical analyses.

10. FA07e_CorrMaps_PerNetwork: This script set will extract summary metrics of voxel-wise connectivity for the 7 Yeo networks. For this purpose we rely on the 3dTcorrmaps generated in step 6). Perhaps a more accurate way to accomplish this is to generate new 3dCorrMaps per network (instead of across the whole GM ribbon), and then generate this network specific maps to extract the per-network summary metrics. This is on the TODO list.

11. FA07f_Voxelwise_Connectivity_Results_7Networks: This notebook shows summary figures for changes in voxel-wise connectivity on a network-by-network basis across flip angles.

12. FA08_ROIbasedConnectivity_Restults: This notebook extract representative time-series per ROI using Nilearn from the fully pre-processed datasets and then computes functional connectivity matrices. Finally, it extract mean within-network connectivity summary metrics and reports them on the form of boxplots.

13. FA09a_BringCleanDataToMNI: This script set brings fully pre-processed datasets into MNI space. Those are later concatenated in 14) to perform seed-based analyses with AFNI instacorr

14. FA09b_BringCleanDataToMNI: This script concatenates fully-preprocessed datasets in MNI space for each flip angle separately.

Instructions for bash scripts

Bash scripts were created to conduct analyses in the biowulf cluster. For that reason, for every step you can find three different scripts:

1. SCRIPT_NAME.sh: This is the main script that perform whatever actions are needed for a given dataset (e.g., a given T1 or EPI scan)

2. SCRIPT_NAME.CreateSwarm.sh: This script will create a swarm file with all the different calls to SCRIPT_NAME.sh so that all datasets are analyzed

3. SCRIPT_NAME.SWARM.sh: This file contains one call to SCRIPT_NAME.sh per dataset. This file is automatically created by SCRIPT_NAME.CreateSwarm.sh

If working on a high computing cluster with swarm, these scripts should work right away. If working on a single computer, you can run SCRIPT_NAME.SWARM.sh, which should analyze each dataset sequentially. (this may take a lot of time).

Many scripts contain a variable PRJDIR that tells the scripts what is the project directory. Make sure to change this variable to the value on your system before running the scripts.

Instructions for Jupyter Notebooks

The python version of this project mostly resides in a series of independent notebooks. Yet these notebooks will use a couple of external files to look project specific information (e.g., project directory, list of subjects, list of flip angles, etc). When you first download this repository, please open the file utils/variables.py and make sure to provide the correct values to PRJDIR (Project Directory), SBJLIST (list of subjects) and FAs (list of flip angles).

In addition, these notebooks rely on a set of python libraries for loading data and plotting data. For your convenience, an yml file with a description of a valid python working environment can be found in the Resources directory. If you have a conda distribution of python on your systems, you can easily create a python virtual envitonment with the necessary libraries using this yml file as a reference.

Additional Files

This repository contains atlas files for the Schaefer Atlas and the Yeo Atlas. Those files were copied from the following publicly available repository (https://github.com/ThomasYeoLab/CBIG)