Add ADNI scripts

adni/10_archive_connectome.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+#SBATCH --account=def-pbellec
+#SBATCH --job-name=conn_archive
+#SBATCH --output=/home/nclarke/scratch/logs/adni_conn_archive.out
+#SBATCH --error=/home/nclarke/scratch/logs/adni_conn_archive.err
+#SBATCH --time=06:00:00
+#SBATCH --cpus-per-task=1
+#SBATCH --mem-per-cpu=8G
+
+
+RAW_PATH="/lustre04/scratch/${USER}/adni_connectomes-0.4.1"
+DATASET_NAME=`basename $RAW_PATH`
+
+ARCHIVE_PATH="/lustre03/nearline/6035398/giga_preprocessing_2/adni_fmriprep-20.2.7lts_1682352545/${DATASET_NAME}"
+
+mkdir -p $ARCHIVE_PATH
+
+cd ${RAW_PATH}
+echo $PWD
+tar -vcf ${ARCHIVE_PATH}/${DATASET_NAME}.tar.gz .

adni/archive_fmriprep.sh → adni/6_archive_fmriprep.sh

@@ -1,9 +1,9 @@
 #!/bin/bash
-#SBATCH --account=rrg-pbellec
+#SBATCH --account=def-pbellec
 #SBATCH --job-name=fmriprep_archive
 #SBATCH --output=/home/nclarke/scratch/logs/adni_fmriprep_archive.out
 #SBATCH --error=/home/nclarke/scratch//logs/adni_fmriprep_archive.out
-#SBATCH --time=48:00:00
+#SBATCH --time=42:00:00
 #SBATCH --cpus-per-task=1
 #SBATCH --mem-per-cpu=8G
 

adni/7_submit_qc_array.sh

@@ -0,0 +1,28 @@
+#!/bin/bash
+#SBATCH --account=def-pbellec
+#SBATCH --job-name=adni_qc
+#SBATCH --output=/lustre04/scratch/nclarke/logs/adni_qc/%x_%A.%a.out
+#SBATCH --error=/lustre04/scratch/nclarke/logs/adni_qc/%x_%A.%a.out
+#SBATCH --time=00:10:00
+#SBATCH --cpus-per-task=1
+#SBATCH --mem=8G
+#SBATCH --array=1-747
+
+module load apptainer
+
+FMRIPREP_DIR=/lustre04/scratch/nclarke/adni_fmriprep-20.2.7lts_1682352545/adni_bids_output_func/fmriprep-20.2.7lts
+GIGA_AUTO_QC_CONTAINER=/home/nclarke/projects/rrg-pbellec/nclarke/giga_preprocess2/giga_auto_qc-0.3.3.simg
+QC_OUTPUT=/lustre04/scratch/nclarke/adni_giga_auto_qc-0.3.3_participant
+participant_labels=/home/nclarke/projects/rrg-pbellec/nclarke/giga_preprocess2/adni/participant_labels.txt # One subject number per line
+
+mkdir -p $QC_OUTPUT
+
+PARTICIPANT_LABEL=$(sed -n "${SLURM_ARRAY_TASK_ID}p" ${participant_labels})
+
+# Create a directory for participant
+PARTICIPANT_OUTPUT="${QC_OUTPUT}/${PARTICIPANT_LABEL}"
+mkdir -p $PARTICIPANT_OUTPUT
+
+echo "Running ${PARTICIPANT_LABEL} QC"
+
+apptainer run --cleanenv -B ${FMRIPREP_DIR}:/inputs -B ${PARTICIPANT_OUTPUT}:/outputs ${GIGA_AUTO_QC_CONTAINER} /inputs /outputs participant --participant_label ${PARTICIPANT_LABEL}

adni/8_archive_qc.sh

@@ -0,0 +1,24 @@
+#!/bin/bash
+#SBATCH --account=def-pbellec
+#SBATCH --job-name=qc_archive
+#SBATCH --output=/home/nclarke/scratch/logs/adni_qc_archive.out
+#SBATCH --error=/home/nclarke/scratch/logs/adni_qc_archive.err
+#SBATCH --time=01:00:00
+#SBATCH --cpus-per-task=1
+
+# Directory to be archived
+RAW_PATH="/lustre04/scratch/${USER}/adni_giga_auto_qc-0.3.3_participant"
+DATASET_NAME=$(basename $RAW_PATH)
+
+# Destination path for the archive
+ARCHIVE_PATH="/lustre03/nearline/6035398/giga_preprocessing_2/adni_fmriprep-20.2.7lts_1682352545"
+
+# Ensure the destination directory exists
+mkdir -p $ARCHIVE_PATH
+
+# Navigate to the parent directory of RAW_PATH
+cd $(dirname $RAW_PATH)
+
+# Tar the directory
+tar -vcf ${ARCHIVE_PATH}/${DATASET_NAME}.tar.gz ${DATASET_NAME}
+

adni/9_connectome_slurm_array.bash

@@ -0,0 +1,47 @@
+#!/bin/bash
+#SBATCH --account=rrg-pbellec
+#SBATCH --output=/lustre04/scratch/nclarke/logs/adni_conn/%x_%A.out
+#SBATCH --error=/lustre04/scratch/nclarke/logs/adni_conn/%x_%A.out
+#SBATCH --cpus-per-task=1
+#SBATCH --array=1-747
+
+GIGA_CONNECTOME_VERSION=0.4.1
+GIGA_CONNECTOME="/home/${USER}/projects/rrg-pbellec/containers/giga_connectome-${GIGA_CONNECTOME_VERSION}.simg"
+FMRIPREP_DIR="/lustre04/scratch/nclarke/adni_fmriprep-20.2.7lts_1682352545/adni_bids_output_func/fmriprep-20.2.7lts"
+CONNECTOME_OUTPUT="/home/nclarke/scratch/adni_connectomes-0.4.1"
+participant_labels="/home/nclarke/projects/rrg-pbellec/nclarke/giga_preprocess2/adni/participant_labels.txt" # One subject number per line
+
+WORKINGDIR="/home/nclarke/scratch/adni_connectomes-0.4.1/working_directory"
+
+module load apptainer
+
+mkdir -p $WORKINGDIR
+
+PARTICIPANT_LABEL=$(sed -n "${SLURM_ARRAY_TASK_ID}p" ${participant_labels})
+PARTICIPANT_OUTPUT="${CONNECTOME_OUTPUT}/${PARTICIPANT_LABEL}"
+
+echo "${FMRIPREP_DIR}"
+if [ -d "${FMRIPREP_DIR}" ]; then
+    mkdir -p ${WORKINGDIR}
+	mkdir -p ${SLURM_TMPDIR}
+    mkdir -p ${CONNECTOME_OUTPUT}
+	mkdir -p ${PARTICIPANT_OUTPUT}
+	echo "Running ${PARTICIPANT_LABEL} connectomes"
+	echo "=========${STRATEGY}========="
+	echo "${ATLAS}"
+	apptainer run \
+		--bind ${FMRIPREP_DIR}:/data/input \
+		--bind ${SLURM_TMPDIR}:/data/output \
+		${GIGA_CONNECTOME} \
+		--atlas ${ATLAS} \
+		--denoise-strategy ${STRATEGY} \
+		${INTRANETWORK_FLAG} \
+		/data/input \
+		/data/output \
+		participant \
+        --participant_label ${PARTICIPANT_LABEL}
+	exitcode=$?  # catch exit code
+	if [ $exitcode -eq 0 ] ; then rsync -rltv --info=progress2 ${SLURM_TMPDIR}/*.h5 ${PARTICIPANT_OUTPUT} ; fi
+else
+    echo "no preprocessed data for ${DATASET}"
+fi

adni/9_submit_connectome_array.sh

@@ -0,0 +1,28 @@
+#!/bin/bash
+#SBATCH --account=rrg-pbellec
+
+DATASET="adni"
+
+STRATEGIES=("acompcor50" "simple" "simple+gsr" "scrubbing.2" "scrubbing.2+gsr" "scrubbing.5" "scrubbing.5+gsr")
+ATLASES=("Schaefer20187Networks" "MIST" "DiFuMo")  
+
+for strategy in "$STRATEGIES"; do
+    for atlas in "${ATLASES[@]}"; do
+            mem=12G
+            time="00:10:00"
+            INTRANETWORK_FLAG="--calculate-intranetwork-average-correlation"
+
+            if [ "${atlas}" == "DiFuMo" ]; then
+                time="00:12:00"
+                mem=14G
+                INTRANETWORK_FLAG=""
+            fi
+
+            echo "Submitting ${atlas} ${strategy}"
+            sbatch \
+                --time=${time} --mem-per-cpu=${mem} \
+                --job-name=${DATASET}_${atlas}_${strategy} \
+                --export=DATASET="${DATASET}",ATLAS="${atlas}",STRATEGY="${strategy}" \
+                ./7_connectome_slurm_array.bash
+    done
+done

adni/README.md

@@ -5,25 +5,40 @@ Scripts for preprocessing ADNI data with fMRIPrep2.2.7lts.
 - Python
 ## Retrieving data
 Data not openly available. Used BIDS-ified dataset on Elm.
-## BIDS errors
+## BIDS validation fixes
 Ran the following scripts to fix errors:
-1. `create_dataset_description.py`
+- `1_create_dataset_description.py`
     -  creates `dataset_description.json` at the root directory with minimal information
     -  BIDS version used is unknown, so specified latest version (see [discussion]( https://neurostars.org/t/what-bids-version-to-use-for-legacy-dataset/25619))
     -  fixes BIDS validation error `code: 57 DATASET_DESCRIPTION_JSON_MISSING`
 
-2. `create_task-rest_bold.py`
+- `2_create_task-rest_bold.py`
     - creates `task-rest_bold.json` at the root directory, detailing bold task
     - fixes BIDS validation error `code: 50 TASK_NAME_MUST_DEFINE`
 
-3. `create_bidsignore.sh`
+- `3_create_bidsignore.sh`
     - creates `.bidsignore` at the root and adds `*T1w_cropped.nii` to ignore these files, which are not needed
     - fixes BIDS validation error `code: 1 - NOT_INCLUDED`
     - TODO: check this also fixes error `code: 63 - SESSION_VALUE_CONTAINS_ILLEGAL_CHARACTER`, it should do
 
-3. `correct_slice_timing.py`
+- `4_correct_slice_timing.py`
     - halves the slice timing for sub-109S4594 (ses-20160502), which appears to be doubled (see [discussion](https://neurostars.org/t/help-with-bids-errors-66-and-75-in-legacy-dataset/25625))
     - fixes BIDS validation error `code: 66 SLICETIMING_VALUES_GREATOR_THAN_REPETITION_TIME`
 
+## Run fMRIPrep
+- `5_generate_slurm_script.sh`
+- `6_archive_fmriprep.sh`
+
+## Run QC
+- `7_submit_qc_array.sh`
+    - run once with `--reindex-bids` flag on one participant to build the database layout (increase wall time)
+    - remove flag and run again with all participants.
+- `8_archive_qc.sh`
+
+## Generate connectomes
+- `9_connectome_slurm_array.bash` and `9_submit_connectome_array.sh`. As above run once with `--reindex-bids` flag and one atlas/strategy. I submitted each atlas/strategy pair individually because I hadn't yet figured out the optimal way to submit them (see e.g. HCP-EP)
+- `10_archive_connectome.sh`
+
 n.b. error `code: 75 - NIFTI_PIXDIM4` affected two subjects. Suggested fix to edit nifti header fields (see [discussion](https://neurostars.org/t/help-with-bids-errors-66-and-75-in-legacy-dataset/25625/2)). One subject would fail anyway, since `Time` field is `0`, so in interests of saving time I have left this for now.
 
+n.b.(b?) around 100 subjects processed without --use-aroma flag. Used detect_timeout.py script to remove.

adni/archive_raw.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+#SBATCH --account=def-pbellec
+#SBATCH --job-name=raw_archive
+#SBATCH --output=/home/nclarke/scratch/logs/adni_raw_archive.out
+#SBATCH --error=/home/nclarke/scratch//logs/adni_raw_archive.err
+#SBATCH --time=12:00:00
+#SBATCH --cpus-per-task=1
+#SBATCH --mem-per-cpu=8G
+
+
+RAW_PATH="/lustre04/scratch/nclarke/adni_bids_output_func"
+DATASET_NAME=`basename $RAW_PATH`
+
+ARCHIVE_PATH="/lustre03/nearline/6035398/giga_preprocessing_2/raw/${DATASET_NAME}"
+
+mkdir -p $ARCHIVE_PATH
+
+cd ${RAW_PATH}
+echo $PWD
+tar -vcf ${ARCHIVE_PATH}/${DATASET_NAME}.tar.gz .

adni/submit_qc_scrub5.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+#SBATCH --account=def-pbellec
+#SBATCH --job-name=qc
+#SBATCH --output=/lustre07/scratch/nclarke/logs/%x_%A.out
+#SBATCH --error=/lustre07/scratch/nclarke/logs/%x_%A.out
+#SBATCH --time=01:00:00
+#SBATCH --cpus-per-task=1
+#SBATCH --mem=12G
+#SBATCH --array=1-747
+
+module load apptainer
+
+FMRIPREP_DIR=/lustre07/scratch/${USER}/adni_bids_output_func/fmriprep-20.2.7lts
+GIGA_AUTO_QC_CONTAINER=/home/${USER}/giga_auto_qc-0.3.3.simg
+QC_OUTPUT=/lustre07/scratch/${USER}/adni_giga_auto_qc-0.3.3_scrub.5
+QC_PARAMS=/home/${USER}/qc_params_scrub5.json
+participant_labels=/home/${USER}/participant_labels.txt # One subject number per line
+
+mkdir -p $QC_OUTPUT
+
+PARTICIPANT_LABEL=$(sed -n "${SLURM_ARRAY_TASK_ID}p" ${participant_labels})
+
+# Create a directory for participant
+PARTICIPANT_OUTPUT="${QC_OUTPUT}/${PARTICIPANT_LABEL}"
+mkdir -p $PARTICIPANT_OUTPUT
+
+echo "Running ${PARTICIPANT_LABEL} QC"
+
+apptainer run --cleanenv -B ${QC_PARAMS} -B ${FMRIPREP_DIR}:/inputs -B ${PARTICIPANT_OUTPUT}:/outputs ${GIGA_AUTO_QC_CONTAINER} /inputs /outputs --quality_control_parameters ${QC_PARAMS} participant --participant_label ${PARTICIPANT_LABEL}