- Launch an ubuntu-based AWS instance with heavy computational power (e.g. m4.16xlarge), see
- SSH to the instance using
ubuntuuser and your private key from AWS
- Clone this repo
git clone https://github.com/mibel/mri_segmentation_dl.git
aws/ami_setup.shscript (it takes approximately 10-20 minutes)
- Create a new EBS volume and mount it as /data
- SCP the dataset to the /data folder
- To launch a nipype based job you should use docker (miykael/nipype_level4 image has been already pulled).
- Also it's possible to launch a job using an (executable) python script from
./scripts/run_cmd.pywhich automatically wrap command and attach volumes:
python scripts/run_cmd.py "echo \$FREESURFER_HOME"