Add a custom Jupyter kernel to JupyterLab

In your JupyterLab session, go to the Launcher and start a new Terminal

Add a Python kernel

Create a conda environment with the required Python version:

module load conda
conda create -n <your-env-name> python=3.7 ipykernel

⚠️

Don’t forget to include the ipykernel package that will be required to use your conda environment in JupyterLab.

Add any other required libraries to the environment

Using conda :

conda activate <your-env-name>
conda install <package>
conda deactivate

Using pip :

conda activate <your-env-name>
pip install <package>
conda deactivate

Install a new kernel for JupyterLab based on your environment

conda activate <your-env-name>
python -m ipykernel install --user --name '<your-env-name>' --display-name "My Python env"
conda deactive

This ia full example to install a new kernel with the latest version of numpy :

Create a conda environment with Python, iPyKernel and Numpy

module load conda
conda create -n my-numpy -c conda-forge python ipykernel numpy

Install the kernel for JupyterLab

conda activate my-numpy
python -m ipykernel install --user --name 'my-numpy' --display-name "Numpy"
conda deactivate

Your Numpy kernel is now available in the launcher

And ready working

Create a conda environment with the required R version

module load conda
conda create -n <your-env-name> r-base=4.2.3 r-irkernel jupyter_client

⚠️

Don’t forget to add the r-irkernel and jupyter_client packages that a required to install your R environment in JupyterLab

Of course, you can add any additionnal package you need.

Activate your conda environment and start a R session:

conda activate <your-env-name>
R

Install your R environment in JupyterLab using the IRKernel library.

IRkernel::installspec(name = '<your-env-name>', displayname = 'My R', user = TRUE)