Uses scripts available at https://github.com/CarolynMcNabb/GUTMIC_pilot_analysis.git
1.1. Create Vanilla VM in nutanix https://rrc.reading.ac.uk:9440/console/#login 1.2. Open VM in NoMachine by pasting IP address from Nutanix 1.3. Open Terminal window
rm -r ~/.mozilla #to clear any open firefox applications
N.B. Conversion of diffusion data to BIDS format may later be updated to use ADWI-BIDS https://arxiv.org/pdf/2103.14485.pdf
BIDS filenames must follow this pattern:
sub-<label>_ses-<label>_task-<label>_acq-<label>_ce-<label>_rec-<label>_dir-<label>_run-<index>_mod-<label>_echo-<index>_flip-<index>_inv-<index>_mt-<label>_part-<label>_recording-<label>
t1_mprage_DC_sag_HCP_256_32ch
| t1_mprage_DC_sag_HCP_256_32ch | MPRAGE structural whole-brain | anat | sub-00x_T1w |
| Resting_ep2d_bold_p2_s4_32ch_TR1500_2.1x2.1_68_32-head-Coil | Resting state fMRI | func | sub-00x_task-rest_bold |
| ep2d_diff_mddw_p2_s4_2mm_2avg_32headcoil | Blip up diffusion MRI | dwi | sub-00x_dir-up_dwi |
| diff_blip_down_mddw_p2_s4_2mm_2avg_32headcoil | Blip down diffusion MRI | dwi | sub-00x_dir-down_dwi |
| gre_field_mapping_32ch | Magnitude and phase images (2 files) | fmap | sub-00x_magnitude1; sub-00x_magnitude2; sub-00x_phasediff |
2.1. Install python 3 and dcm2bids In terminal, type:
sudo apt install python3.7
sudo apt install python-pip
pip install dcm2bids
2.2. Set up environment for dcm2bids
sudo apt install gedit
gedit environment.yml
Paste in the following:
name: dcm2bids
channels:
- conda-forge
dependencies:
- dcm2niix
- dcm2bids
- pip
Add path:
PATH=$PATH:/opt/anaconda/bin/
export PATH
Create conda environment and activate:
conda env create -f environment.yml --prefix /storage/shared/research/cinn/2018/GUTMIC/CM_scripts/conda_env/
source activate /storage/shared/research/cinn/2018/GUTMIC/CM_scripts/conda_env
cd /storage/shared/research/cinn/2018/GUTMIC/
2.3. Convert raw data to BIDS format with dcm2bids First, test that dcm2bids is working by typing
dcm2bids --help
Next, make sure dcm2bids_config.json contains the following:
{
"descriptions": [
{
"dataType": "anat",
"modalityLabel": "T1w",
"criteria": {
"ImageType": ["ORIGINAL", "PRIMARY", "M", "NORM", "DIS2D"],
"SequenceName": "*tfl3d1_16ns*",
"SeriesDescription": "t1_mprage*"
}
},
{
"dataType": "func",
"modalityLabel": "bold",
"customLabels": "task-rest",
"criteria": {
"SeriesDescription": "*Resting_ep2d_bold*"
},
"sidecarChanges": {
"TaskName": "rest"
}
},
{
"dataType": "fmap",
"modalityLabel": "magnitude1",
"criteria": {
"SeriesDescription": "gre_field_mapping_32ch", "SidecarFilename": "*_e1.*"
},
"intendedFor": 0
},
{
"dataType": "fmap",
"modalityLabel": "magnitude2",
"criteria": {
"SeriesDescription": "gre_field_mapping_32ch", "SidecarFilename": "*_e2.*"
},
"intendedFor": 0
},
{
"dataType": "fmap",
"modalityLabel": "phasediff",
"criteria": {
"SidecarFilename": "*_e2_ph*"
},
"intendedFor": 0
},
{
"dataType": "dwi",
"modalityLabel": "dwi",
"customLabels": "dir-up",
"criteria": {
"SequenceName": "*ep_b0",
"ImageType": ["ORIGINAL", "PRIMARY", "DIFFUSION", "NONE", "ND", "NORM", "MOSAIC"],
"SeriesDescription": "*ep2d_diff*"
}
},
{
"dataType": "dwi",
"modalityLabel": "dwi",
"customLabels": "dir-down",
"criteria": {
"SeriesDescription": "*diff_blip_down*"
}
}
]
}
Then type into terminal (within the conda environment):
2.0_dcm2bids.sh
Alternatively, type:
cd /storage/shared/research/cinn/2018/GUTMIC/CM_MRS/
dcm2bids -d /storage/shared/research/cinn/2018/GUTMIC/raw/GUTMIC_002/ -p 002 -c /storage/shared/research/cinn/2018/GUTMIC/CM_scripts/code/dcm2bids_config.json
And Repeat for every subject
Add modality agnostic file (dataset_description.json). See bids specification for details.
2.4. Validate BIDS format using BIDS validator https://pypi.org/project/bids-validator In terminal window (not in python) type:
pip install bids_validator
python3.7 2.1_validate_bids.py
OR go to https://bids-standard.github.io/bids-validator/ and import bids parent folder (folder containing all subjects)
2.5 Finalise folder structure, including making directories (mkdir) for analysis:
/storage/shared/research/cinn/2018/GUTMIC/
func_diff/
derivatives/
TBSS/
preprocessed/
sub_001/
…
analysis/
fMRIprep/
sub_001/
anat/
dwi/
fmap/
func/
sub_002/
...
Uses FSL 6.0.1 on an ubuntu MATE 16.04 operating system (8GB). Also uses python 3.8.3 for making directories in the first step (3.0). 3.0. Make the fMRI derivatives directory and subject subdirectories for use during the analysis. In a python console, type:
3.0_makedirs.py
3.1 Next, perform brain extraction of T1w scan for each participant using FSL's bet function. In the Ubuntu terminal, type:
3.1_brainextraction.sh
If brain extraction was unsuitable (you can check which subjects you were unhappy with in the brain_extraction_checks.txt file), you can run 3.1.1_betcleanup.sh to improve the extraction for those participants. Before running, you will need to open the script and edit the code to include only those subjects that need additional clean up (line 17). Once this is done, save the script and in the Ubuntu terminal, type:
3.1.1_betcleanup.sh
3.2. Create a B0 fieldmap for bias field correction during registration in FEAT. In the Ubuntu terminal, type:
3.2_fieldmap.sh
3.3. Run the first FEAT step (FEATpreproc) to produce independent components that will be used for motion correction. For this step, you need to have put the FEATpreproc.fsf file in a folder that you can access and amend the path to that folder in the 3.3_FEATpreproc.sh script. After doing that, in the Ubuntu terminal, type:
3.3_FEATpreproc.sh
3.4. To save on space - I will not perform additional motion correction with FIX for this dataset (I have deleted the melodic.ica folders from each of the subjects' directories). Instead, I'll just jump into the group ICA.
3.5. Warp functional data from subject space into standard space. In Ubuntu terminal window, type:
3.5_warp2std.sh
3.6. Smooth the functional data (now in standard space) using a 5 mm Gausian kernel and then downsample to 2mm (currently in 1mm space) to increase speed of MELODIC. In ubuntu terminal, type:
3.6_smoothdownsample.sh
In addition, create a GLM containing the faecal GABA, Glutamate and Glutamine levels from LCMS. In R, run the following script:
3.6_GLMSetup.R
After doing this, open the GLM gui in FSL. In the ubuntu terminal, type:
Glm
In Glm setup window:
In General Linear Model window:
In Higher-level model - paste window:
In General Linear Model window:
|
3.7. Create group-level independent components using FSL's MELODIC. In the ubuntu terminal, type:
3.7_melodic.sh
3.8. Run dual regression in FSL to estimate a "version" of each of the group-level spatial maps for each subject. Dual regression regresses the group-spatial-maps into each subject's 4D dataset to give a set of timecourses (stage 1) and then regresses those timecourses into the same 4D dataset to get a subject-specific set of spatial maps (stage 2). In the ubuntu terminal, type:
3.8_dualregression.sh
3.9. Dual regression will not perform an ANOVA and only t-tests can be viewed in the dual regression output directory at present. To run F-tests you need to run randomise on the dual regression output. Although this isn't necessary for the GUTMIC pilot, it may be required for the GutBrainGABA study so I include it here for completeness. In the ubuntu terminal window, type:
3.9_randomise.sh
Uses FSL 6.0.1 on an ubuntu MATE 16.04 operating system (8GB) 4.0. FSL prefers bvec and bval files to be named bvecs and bvals for analysis purposes. First step is to link the bval and bvec files in the BIDS folders to the preprocessing directory. In the terminal window, type:
4.0_link_bvs.sh
4.1. Create a file called b0_images.nii.gz - a 4D image file containing eight b=0 volumes that have been acquired using different phase encoding directions (4 times P>>A; -j and 4 times A>>P; j) so the off-resonance field distortion is different in the different volumes. In the terminal window, type:
4.1_join_b0s.sh
4.2. Estimate the susceptibility induced off-resonance field using FSL's topup. In the terminal window, type:
4.2_topup.sh
Notes: an acq_param.txt file for the b0_images, containing the acquisition parameters and readout time, as well as the b02b0.cnf file, and index file are included in the github repository and should be stored in the “code_path” folder
4.3. Generate a brain mask of the non-distorted hifi_b0 image. In terminal window, type:
4.3_mask.sh
Notes: check each subject’s mask to make sure the automatic brain extraction worked well - lower opacity to view both mask and brain images together..
fsleyes hifi_b0.nii.gz hifi_b0_brain_mask.nii.gz -cm Yellow
If the mask does not cover all areas of the brain, try re-creating the mask using:
bet hifi_b0.nii.gz hifi_b0_brain -f 0.4 -m
-f will make the mask bigger (bet will be less stringent) This step was carried out for: sub_008, sub_015
4.4. Perform correction of eddy current-induced distortions and subject movements using FSL's eddy_openmp. In terminal window, type:
4.4_eddy.sh
Note: slice-to-volume correction was not used for this analysis as it is unavailable with the CPU version of eddy and also is not expected to provide much of an improvement in healthy adult volunteers (such as those used in the current study) Manually check each subject’s eddy_corrected_data.nii.gz file using FSL’s fsleyes before proceeding.
4.5. Get mean motion parameters for each participant, including mean motion from first volume and mean motion from previous volume. In the terminal window, type:
4.5_motion.sh
4.6. Extract outlier information and descriptive stats from motion data. These data need to be reported along with any results. See doi: 10.1016/j.neuroimage.2018.02.041. In addition, use these values, along with gut diversity measures, to create csv files that can be copied into FSL’s GLM gui to create .mat and .con files for use in FSL’s Randomise. In R, run the following scripts:
4.6_outliers.R
4.6_GLMsetup.R
Notes: GLM files can be found in the GLMs folder in the github directory https://github.com/CarolynMcNabb/GUTMIC_pilot_analysis.git
4.7. Fit diffusion tensors to the eddy-corrected data. In the terminal window, type:
4.7_dtifit.sh
4.8. Run tract-based spatial statistics on the data. In the terminal window, type:
4.8_TBSS.sh
Notes: this script runs several different commands and requires user interaction to move onto the next stage of analysis. Read the instructions in the terminal and respond when requested.
4.9 Run nonparametric permutation inference on the TBSS output using the GLM files available in the GLMs directory (created using FSL's Glm tool). In the terminal window, type:
4.9_randomise.sh
Notes: at the end of the analysis, check the inferential_stats.txt file in the derivatives/TBSS/analysis/stats/ directory to see if any voxels in the t contrast are statistically significant. If the values in the third column are above .95, this indicates that the t value reached significance (presuming that you do not need to correct for multiple comparisons). The script will also open FSLeyes so you can view the significant voxels in the brain.