Setup

1. Launch an ubuntu-based AWS instance with heavy computational power (e.g. m4.16xlarge), see aws/aws_config.json for example.
2. SSH to the instance using ubuntu user and your private key from AWS
3. Clone this repo git clone https://github.com/mibel/mri_segmentation_dl.git
4. Launch aws/ami_setup.sh script (it takes approximately 10-20 minutes)
5. Create a new EBS volume and mount it as /data
6. SCP the dataset to the /data folder

Usage

1. To launch a nipype based job you should use docker (miykael/nipype_level4 image has been already pulled).
2. Also it's possible to launch a job using an (executable) python script from ./scripts/run_cmd.py which automatically wrap command and attach volumes:

python scripts/run_cmd.py "echo \$FREESURFER_HOME"

See also run_cmd.py --help