Abstract

Project Lead

Taylor Salo

Faculty Lead

Theodore D. Satterthwaite

Analytic Replicator

Tien Tong

Collaborators

List to be collected

Project Start Date

August 2023

Current Project Status

Manuscript in preparation

Datasets

Github Repository

https://github.com/PennLINC/mebold-trt

Slack Channel:

#mebold-abcd-compare

Cubic Project Directory Overview

/cbica/projects/executive_function/mebold_trt/

CODE DOCUMENTATION

All project analyses are described below along with the corresponding code on Github. The following outline describes the order of the analytic workflow:

0. Download data from FlyWheel and curate for OpenNeuro 1. Download from OpenNeuro 2. Run processing scripts

In each of the scripts, there is a variable REPRO_DIR or repro_dir that needs to be set to the directory where you’re running the replication from.

0. Download data from FlyWheel

The steps to create the initial BIDS data can be found in processing_code/curation. Briefly, the steps were to download the dicom tarballs from Flywheel and run heudiconv. Because no-RF images are included in the dicoms, these were split into separate files. The anatomical scans were refaced using AFNI. Finally, the events files were converted to BIDS format.

The validated BIDS data were uploaded to OpenNeuro. This is where the replication process can begin - with the datalad cloned BIDS data from OpenNeuro.

1. Download from OpenNeuro

As the executivefunction user on cubic we cloned the data from OpenNeudo using

micromamba activate babs
pip install babs==0.5.4

cd /cbica/projects/executive_function/mebold_trt
datalad install https://github.com/OpenNeuroDatasets/ds005250.git

cd /cbica/projects/executive_function/mebold_trt/ds005250
datalad get -J 10 .

Configure the remote S3 repository to have the highest cost, ensuring that files from the local repo are always preferred and used first. Edit the .git/config:

[remote "s3-PUBLIC"]
	annex-s3 = true
	annex-uuid = 683c9a58-3e80-42a2-b641-01d5becfcdda
	annex-cost = 500.0 # add this line

2. Run processing scripts

2.1. MRIQC v24.0.2

Pull the Apptainer image and create a DataLad dataset containing the image

apptainer_path=/cbica/projects/executive_function/mebold_trt/software/apptainer

apptainer build \
    ${apptainer_path}/mriqc-24-0-2.sif \
    docker://nipreps/mriqc:24.0.2


apptainerDS_path=/cbica/projects/executive_function/mebold_trt/software/apptainer-ds

datalad create -D "Create mriqc-24-0-2 DataLad dataset" ${apptainerDS_path}/mriqc-24-0-2-ds
cd ${apptainerDS_path}/mriqc-24-0-2-ds
datalad containers-add \
    --url ${apptainer_path}/mriqc-24-0-2.sif \
    mriqc-24-0-2

Run MRIQC with BABS using the following scripts:

# create the BABS project
bash babs_mriqc_init.sh

# Run BABS
cd /cbica/projects/executive_function/mebold_trt/derivatives/mriqc_babs_project
# Run `babs submit` and `babs status`
# Run `babs merge` when all jobs finish successfully

# Create an ephemeral clone + unzip outputs
bash babs_mriqc_finish.sh

2.2. NORDIC (v0.0.1) + fMRIPrep (v25.2.3)

Pull the Apptainer image and create a DataLad dataset containing the image

apptainer_path=/cbica/projects/executive_function/mebold_trt/software/apptainer

apptainer build \
    ${apptainer_path}/fmriprep-25-2-3.sif \
    docker://nipreps/fmriprep:25.2.3


cd /cbica/projects/executive_function/mebold_trt/software
git clone https://github.com/nipreps/NORDIC_Raw.git
cd NORDIC_Raw
git checkout executable
apptainer build ${apptainer_path}/nordic-0-0-1.sif Apptainer.def

###############################################################################

apptainerDS_path=/cbica/projects/executive_function/mebold_trt/software/apptainer-ds

datalad create -D "Create nordic-0-0-1 and fmriprep-25-2-3 DataLad dataset" ${apptainerDS_path}/nordic-fmriprep-ds
cd ${apptainerDS_path}/nordic-fmriprep-ds

datalad containers-add \
    --url ${apptainer_path}/fmriprep-25-2-3.sif \
    fmriprep-25-2-3

datalad containers-add \
    --url ${apptainer_path}/nordic-0-0-1.sif \
    nordic-0-0-1

Run NORDIC+fMRIPrep with BABS using the following scripts:

Create the BABS project by running bash babs_nordic_fmriprep_init.sh. Once the project is create, add fmriprep_filter.json and nordic_filter.json to nordic_fmriprep_babs_project/analysis/code, and push the filter files.

cp \
  /cbica/projects/executive_function/mebold_trt/derivatives/code/fmriprep_filter.json \
  /cbica/projects/executive_function/mebold_trt/derivatives/nordic_fmriprep_babs_project/analysis/code
cp \
  /cbica/projects/executive_function/mebold_trt/derivatives/code/nordic_filter.json \
  /cbica/projects/executive_function/mebold_trt/derivatives/nordic_fmriprep_babs_project/analysis/code

cd /cbica/projects/executive_function/mebold_trt/derivatives/nordic_fmriprep_babs_project
babs sync-code -m "add nordic and fmriprep filter files"

###############################################################################

# Run BABS
cd /cbica/projects/executive_function/mebold_trt/derivatives/nordic_fmriprep_babs_project
# Run `babs submit` and `babs status`
# Run `babs merge` when all jobs finish successfully

# Create an ephemeral clone + unzip outputs
bash babs_nordic_fmriprep_finish.sh

2.3. tedana

Run tedana using the following scripts:

# create subject-pair tsv
python generate_sub_ses_pairs.py

# Run tedana
cd /cbica/projects/executive_function/mebold_trt/github/parker/processing
sbatch run_tedana.sbatch

2.4. XCP-D v0.13.0

Pull the Apptainer image and create a DataLad dataset containing the image

apptainer_path=/cbica/projects/executive_function/mebold_trt/software/apptainer

apptainer build \
    ${apptainer_path}/xcpd-0-13-0.sif \
    docker://pennlinc/xcp_d:0.13.0


###############################################################################

apptainerDS_path=/cbica/projects/executive_function/mebold_trt/software/apptainer-ds

datalad create -D "Create xcpd-0-13-0 DataLad dataset" ${apptainerDS_path}/xcpd-0-13-0-ds
cd ${apptainerDS_path}/xcpd-0-13-0-ds
datalad containers-add \
    --url ${apptainer_path}/xcpd-0-13-0.sif \
    xcpd-0-13-0

Run XCP-D with BABS using the following scripts:

For single-echo:

# create the BABS project
bash babs_xcpd_SE_init.sh

# add filter file
cp \
	/cbica/projects/executive_function/mebold_trt/derivatives/code/xcpd_SE_filter.json \
	/cbica/projects/executive_function/mebold_trt/derivatives/xcpd_SE_babs_project/analysis/code
cd /cbica/projects/executive_function/mebold_trt/derivatives/xcpd_SE_babs_project
babs sync-code -m "Add xcpd SE filter file"

# Run BABS
cd /cbica/projects/executive_function/mebold_trt/derivatives/xcpd_SE_babs_project
# Run `babs submit` and `babs status`
# Run `babs merge` when all jobs finish successfully

# Create an ephemeral clone + unzip outputs
bash babs_xcpd_SE_finish.sh

For multi-echo:

First, need to nano /cbica/projects/executive_function/mebold_trt/derivatives/tedana/dataset_description.json

{
    "Name": "tedana",
    "BIDSVersion": "1.5.0",
    "DatasetType": "derivative",
    "GeneratedBy": [{"Name": "tedana"}]
}

Then, save tedana output as a datalad dataset

cd /cbica/projects/executive_function/mebold_trt/derivatives/
datalad create -c text2git tedana_dl
cp -r tedana/* tedana_dl
datalad save \
	-d /cbica/projects/executive_function/mebold_trt/derivatives/tedana_dl \
	-m "checked dataset into datalad"

Lastly, use BABS to run XCP-D

# create the BABS project
bash babs_xcpd_ME_init.sh

# add filter file and tedana_xcp_d_config
cp \
	/cbica/projects/executive_function/mebold_trt/derivatives/code/xcpd_ME_filter.json \
	/cbica/projects/executive_function/mebold_trt/derivatives/xcpd_ME_babs_project/analysis/code
cp \
	/cbica/projects/executive_function/mebold_trt/derivatives/code/tedana_xcp_d_config.yml \
	/cbica/projects/executive_function/mebold_trt/derivatives/xcpd_ME_babs_project/analysis/code
cd /cbica/projects/executive_function/mebold_trt/derivatives/xcpd_ME_babs_project
babs sync-code -m "add xcpd ME filter file and tedana_xcp_d_config"

# Run BABS
cd /cbica/projects/executive_function/mebold_trt/derivatives/xcpd_ME_babs_project
# Run `babs submit` and `babs status`
# Run `babs merge` when all jobs finish successfully

# Create an ephemeral clone + unzip outputs
bash babs_xcpd_ME_finish.sh

2.5. Fractal n-back GLMs with Nilearn

micromamba activate meboldtrt

cd /cbica/projects/executive_function/mebold_trt/github/parker/analysis

# Run first-level (run-wise) GLMs with tedana confounds
python run_nback_first_level_rtdur.py
# Run second-level (group-level) GLMs from outputs from first-level GLMs
python run_nback_second_level_rtdur.py
# Plot second-level results
python plot_nback_second_level.py

Multi Echo BOLD Comparison Pilot Replication Guide

Project Information and Reproducibility Guide

Abstract

Project Lead

Faculty Lead

Analytic Replicator

Collaborators

Project Start Date

Current Project Status

Datasets

Github Repository

Slack Channel:

Cubic Project Directory Overview

CODE DOCUMENTATION

0. Download data from FlyWheel

1. Download from OpenNeuro

2. Run processing scripts

2.1. MRIQC v24.0.2

2.2. NORDIC (v0.0.1) + fMRIPrep (v25.2.3)

2.3. tedana

2.4. XCP-D v0.13.0

2.5. Fractal n-back GLMs with Nilearn