Runs fmri pipeline using singularity image
Sample_input_data is provided in this repo.
Optional parameters info: The optional params info is detailed in fmri optional params description.xlsx
json input file is required.
Ex json input to run_script.py:
{"data":[["/home/vagrant/input1","sub1","sess1"],["/home/vagrant/input2","sub2","sess2"]], "options":{
"reorient_params_x_mm":0,
"reorient_params_y_mm":0,
"reorient_params_z_mm":0,
"reorient_params_pitch":0,
"reorient_params_roll":0,
"reorient_params_yaw":0,
"realign_fwhm":8,
"realign_interp":2,
"realign_quality":1,
"realign_register_to_mean":true,
"realign_separation":4,
"realign_wrap": [0,0,0],
"realign_write_interp": 4,
"realign_write_mask": true,
"realign_write_which": [2, 1],
"realign_write_wrap": [0,0,0],
"slicetime_ref_slice":16,
"normalize_affine_regularization_type":"mni",
"normalize_write_bounding_box": [[-78, -112, -70],[78, 76, 85]],
"normalize_write_interp": 1,
"normalize_write_voxel_sizes": [3,3,3],
"smoothing_x_mm":10.0,"smoothing_y_mm":10.0,"smoothing_z_mm":10.0,"smoothing_implicit_masking":false
}}
Note: trendscenter/fmri_web_image hosted on dockerhub is private. So contact me with your docker ID so that I can give you access to it.
You can run the code in foll. ways:
- Build dockerfile file from here docker build -t fmri_image_name Dir_containing_Dockerfile
Sometimes you may create a docker container on your local machine and when you move it elsewhere you may not have permissions to run somethings inside that container. So go inside the container and run "chmod -R 755 /" or equivalent to make sure your required software is executable.
To change the cache dir location , incase you dont have permissions to write to the default in bash: export SINGULARITY_CACHEDIR="/user/container"
same with writing to default /tmp folder export SINGULARITY_TMPDIR="/user/container/tmp"
then run singularity build fmri.img trendscenter/fmri_web_image if you want to create the fmri.img as writable image for editing , then : singularity build --sandbox fmri.img docker://trendscenter/fmri_web_image or if you have permissions to use --writable then: singularity build --writable fmri.img docker://trendscenter/fmri_web_image
This is what I used and it works fine.
Sometimes due to changes in open source code of packages being used in this container, things may not work as expected. If you want to use the image that I already built then proceed to step 2)
- use trendscenter/fmri_web_image located on dockerhub
To change the cache dir location , incase you dont have permissions to write to the default in bash: export SINGULARITY_CACHEDIR="/export/user/container"
same with writing to default /tmp folder export SINGULARITY_TMPDIR="/user/container/tmp"
singularity build fmri.img docker://trendscenter/fmri_web_image
if you want to create the fmri.img as writable image for editing , then : singularity build --sandbox fmri.img docker://trendscenter/fmri_web_image or if you have permissions to use --writable then: singularity build --writable fmri.img docker://trendscenter/fmri_web_image
This is what I used and it works fine.
- Once downloaded to local machine:
3.a) You can use run_script.py and supply required args to have the script call singularity image and run the pipeline or do it manually as in step 3.b)
python run_script.py -h usage: run_script.py [-h] -json JSON -o OUTPUT -tmp TMP -image_path IMAGE_PATH [-g SLURM] [-type {nifti,dicoms}]
Run fmri pipeline on mprage T1w dicoms
optional arguments: -h, --help show this help message and exit
required arguments: -json JSON, --json JSON
-o OUTPUT, --output OUTPUT Existing output directory to store vbm outputs -tmp TMP Nipype tmp directory, it can be same as output directory -image_path IMAGE_PATH Enter full path to singularity vbm image name
optional arguments: -g SLURM, --slurm SLURM Use slurm, True or False, default=False -type {nifti,dicoms}, --data_type {nifti,dicoms} input data type. Ex:'nifti','dicoms'
python run_script.py -image_path /container/gsu_fmri.img/ -i /container/input -o /container/output -json /container/json_data/input.json -tmp /output -sub sub1 -sess sess1 args passed are: Namespace(data_type='nifti', image_path=['/container/gsu_fmri.img/'], input=['/container/input'], output=['/container/output'], sess=['sess1'], slurm=False, smooth=None, sub=['sub1'], tmp=['/container/output']) Executed command is: singularity exec --contain --workdir /container/output -B /container/input:/data,/container/output:/out /container/gsu_fmri.img/ python3 /computation/run_fmri.py -sub sub1 -sess sess1 -json '/json_data/input.json'
fmri pipeline is running... fmri output directory: /out/sub1/sess1/anat fmri preprocessing completed.
3.b) Run singularity singularity exec --contain --workdir NIPYPE_TMP_DIR -B INPUT_DIR:/data,OUTPUT_DIR:/out fmri.img python3 /computation/run_fmri.py -sub SUBJECT_ID -sess SESSION_ID -json '/json_data/input.json'
-c/--contain This option disables the sharing of filesystems on
your host (e.g. /dev, $HOME and /tmp).
--workdir stores tmp files that singulairty image will write , in this case nipype(package that implements fmri pipeline) writes to this. Nipype will try to write to host /tmp or var_tmp and will have permission issue. Hence usage of this option.
-B Binds path on host to container
Variables:
NIPYPE_TMP_DIR : Directory to store nipype tmp files, can be anywhere that the image has write access to. Could be stored in sessions output dir in the logs as well.
INPUT_DIR : Full path to directory where input data is located : Data can be a nifti file or dicoms
The script assumes there is only one nifti file
OUTPUT_DIR : Directory where outputs will be written to
Required arguments:
Each input scan is from sepecific subject and sessionthat The foll. arguemnts are used to create output directories using these args
SUBJECT_ID : subject id
SESSION_ID : session id
json : full path to accessible json file for the singularity image
Sample output data: ls output/fmri_outputs/ outputs_description.txt quality_control_readme.txt sub1
ls output/fmri_outputs/sub1/sess1/func/fmri_spm12
QC_Framewise_displacement.txt meansub2.nii rsub2.nii wasub2.nii asub2.nii rp_sub2.txt swasub2.nii