vitessce-python

Python API and Jupyter widget facilitating interactive visualization of spatial single-cell data with Vitessce.

Installation

To install with pip:

$ pip install 'vitessce[all]'

Getting started

Explore our example notebooks. These contain demos of different use cases and integrations with single-cell data analysis packages.

Development

For a development installation (requires NodeJS and NPM),

$ git clone https://github.com/vitessce/vitessce-python.git
$ cd vitessce-python
$ uv sync --extra dev --extra docs --extra all

Alternatively, use conda:

$ conda create -n vitessce-dev python=3.12
$ conda activate vitessce-dev
$ pip install -e ".[dev,docs,all]"

Troubleshooting SpatialQuery installation on macOS:

$ # Use the macOS clang instead of from homebrew.
$ CC=/usr/bin/clang CXX=/usr/bin/clang++ pip install -e ".[dev,docs,all,sq]"

Linting and testing

uv run flake8
uv run pytest

Formatting

One file:

uv run autopep8 --in-place --aggressive ./path/to/file.py

All .py files:

find . -name '*.py' -exec uv run autopep8 --in-place --aggressive '{}' \;

Documentation

uv run make html

Deployment

To deploy a new version, increment the version of the Python package in pyproject.toml.

Then, when you push or merge the code with the incremented versions to main, the GitHub Action deploy.yml workflow will build and push the package to PyPI.

Troubleshooting

Check JavaScript console

Check the JavaScript console in the web browser for error messages. To do so, shift+right click in Jupyter, then click Inspect element. You may need to enable your browser's develop mode in its preferences for this option to appear when you right-click.

Check your Python environment

Check that you have activated the correct conda or other virtual environment as you expect. When sharing errors, it can be helpful to share the current versions of packages in the environment, e.g., by conda list or pip list.

Restart Kernel and Clear All Outputs, then refresh the browser tab

To ensure that older widget JavaScript output is not causing conflicts with newer JavaScript output (which may persist via notebook outputs even without running notebook cells), clear old outputs by Kernel -> Restart Kernel and Clear All Outputs. Then refresh the browser tab (containing Jupyter) to ensure all JavaScript outputs are fresh.

This is also typically required when running !pip install vitessce[all] from within a Jupyter notebook cell in the same session as running vc.widget() for the first time. The Kernel restart/Output clearing is required when a new Jupyter widget is installed in the environment, otherwise Jupyter will not yet recognize the newly-installed widget (as the notebook was opened with a different set of Jupyter widgets installed, and widget discovery is performed only at certain stages in the notebook lifecycle).

Check the widget configuration

If the widget renders successfully, you can get its current configuration by

vw = vc.widget()
vw

vw.config

Usage when Jupyter is running on a remote machine

If Jupyter is running on a remote machine, then use proxy=True. You may need to specify host_name as well (widget should be able to detect this but the plain display cannot).

vw = vc.widget(proxy=True)
vw

or

vc.display(proxy=True, host_name="http://localhost:8888")

For more information see the data location options documentation.

Ensure that Vitessce and Jupyter are installed in the same environment

With the environment activated, run pip list on the command line:

Package                       Version
----------------------------- -----------
...
jupyterlab                    3.x.x
...
vitessce                      3.x.x

Resources

• ipywidget docs: Building a Custom Widget
• ipywidget docs: Low Level Widget Tutorial
• ipywidget example: ipyleaflet
• ipywidget example (with React): ipymaterialui
• ipywidget example (with React): higlass-python
• ipywidget cookiecutter
• Sphinx: Getting Started
• Read the Docs Sphinx Theme
• jupyter server proxy

Getting/Offering Help

If you have a specific bug or feature request, please feel free to open an issue. Otherwise our discussions section is a great place to get help or offer it. If you aren't sure if something is a bug or not, don't have all the reproduction steps, or just have a general question, feel free to open a discussion post.