Installing Qiime2 extension under JupyterLab 4.x.x


I'm encountering challenges integrating QIIME 2 (version 2024.2, installed via conda) with JupyterHub, which is running as a system service in a separate conda environment. My goal is to use QIIME 2's functionalities within JupyterHub, with QIIME 2 installed in its standalone environment and registered with ipywidgets in JupyterHub. Unfortunately I can't install the qiime2 widget for the JupyterHub, as it has JupyterLab version 4.1.2 and it uses ExtensionManager that uses PyPI as source of the extensions.

How can I install it differently?

Hey @Michal_Kowalski,

I do not believe there will be any good way to get the QIIME 2 widget to work in that environment.

Sorry. You aren't necessarily missing out on too much, as the most it does is show you the visualization inline.

Just to confirm, you are still able to use the QIIME 2 environment as a kernel inside JupyterHub, right?

Partially. I have had problems with DADA2 to find the right path.
Overall in the invocation of the process, eithe if it is being run by subprocess or system, the conda run -n qiime2-amplicon-2024.2 seems like possible solution :slight_smile:

Oh I suppose that issue would make sense. I'm unfamiliar if the Jupyter kernel for conda will activate the environment in a way that causes our hooks run (which unset R and Python local packages via an environment variable).

As a slightly more convenient in-between, you could try getting the kernel to set these environment variables to either nothing or some irrelevant value:

I think that ought to help with the R path issue.

1 Like