JupyterLab plugin that brings the omicverse-web DataFrame and AnnData preview ideas into a standalone package.

It provides:

• Rich output renderers for pandas.DataFrame, pandas.Series, and anndata.AnnData
• A JupyterLab command that opens an OmicVerse Notebook panel for inspecting kernel variables by name
• Automatic formatter activation for the active notebook or console kernel when the JupyterLab frontend extension is loaded
• Color-coded DataFrame columns, sticky headers, compact shape cards, and AnnData slot summaries

This package intentionally focuses on the display layer:

• kernel-side Python helpers turn variables into JSON preview payloads
• JupyterLab frontend plugins render those payloads and provide a lightweight inspector UI

It does not depend on the omicverse-web Flask APIs.

Development install:

cd omicverse-notebook
pip install -e .

For an end-user install, build a wheel first so the prebuilt labextension is bundled:

cd omicverse-notebook
jlpm install
jlpm build:prod
pip install .

If the frontend extension is loaded correctly, it will try to enable formatters automatically for the active notebook or console kernel.

Manual fallback:

%load_ext omicverse_notebook

Or:

from omicverse_notebook import enable_formatters
enable_formatters()

After that, displaying a DataFrame or AnnData object in a notebook cell will use the OmicVerse renderers.

Open Command Palette and run:

OmicVerse Notebook: Open

If rich output still does not appear in an already-running kernel, run:

OmicVerse Notebook: Enable Kernel Formatters

Then inspect variables like:

df
adata
adata.obs
adata.layers["counts"]
adata.obsm["X_umap"]

• The inspector executes a small helper snippet in the current kernel and expects the Python package to be installed in that kernel environment.
• AnnData is optional on install, but AnnData previews only activate if anndata is available in the kernel.