You can create custom kernels from an Anaconda environment or a Singularity container. In both cases you’ll need to install the ipykernel package.

Jupyter kernel based on Anaconda environment

To create a custom kernel of the Anaconda environment myenv that uses Python 3.7:

module load anaconda/2019.10-py3.7
conda create -n myenv python=3.7 ipykernel <other_packages>
source activate myenv
python -m ipykernel install --user --name myenv --display-name "My Env"


  • You can customize the display name for your kernel. It is shown when you hover over a tile in JupyterLab. If you do not specify a display name, the default Python [conda env:<ENV_NAME>] will be shown.
  • For more information on Anaconda, please visit here.

Jupyter kernel based on Singularity container

For this to work, the ipykernel Python package must be installed within the Singularity container. To create a Jupyter kernel for the container, you can either use our automated script jkrollout or do it manually.

Automated script

Replace /path/to/sif with the actual image name or path:

jkrollout /path/to/sif "My kernel"

If GPU is supported:

jkrollout /path/to/sif "My kernel" gpu


Custom kernels are stored under ~/.local/share/jupyter/kernels. If this directory does not already exist, run

mkdir -p ~/.local/share/jupyter/kernels

Next, cd into it and create a directory for your specific kernel, e.g. mykernel:

mkdir mykernel

Create two files in that directory, kernel.json and (The former must be exactly kernel.json. The latter can be customized as long as you are consistent.)


 "argv": [
 "display_name": "My kernel",
 "language": "python"

(Remember to replace your_id with your user ID.)

module load singularity
singularity exec /path/to/singularity/image python -m ipykernel $@

(Remember to use the actual path of your Singularity image.)

If the container has GPU support, add a --nv flag in the last line:

singularity exec --nv /path/to/singularity/image python -m ipykernel $@

Change into an executable:

chmod +x

You will see your custom kernel “My kernel” next time you use JupyterLab.