Clustergrammer2 is an interactive heatmap Jupyter widget built using the widget-ts-cookiecutter library and the WebGL library regl. Clustergrammer2 is built to help researchers interactively explore single cell data (e.g. scRNA-seq). Please see Case Studies and Tutorials for examples.

Clustergrammer2 Examples

Basic Example of Running Clustergrammer2 on MyBinder

The above notebook shows how Clustergrammer2 can be used to load a small dataset and visualize a large random DataFrame. By running the notebook on MyBinder using Jupyter Lab it can also be used to visualize a user uploaded dataset. Please see the video tutorial above for more information.

For additional examples and tutorials please see:

Case Studies and Tutorials
Clustergrammer2-Notebooks GitHub repository

2,700 PBMC scRNA-seq

Single cell RNA-seq (scRNA-seq) is a powerful method to interrogate gene expression across thousands of single cells. This method produces thousands of measurements (single cells) across thousands of dimensions (genes). This notebook uses Clustergrammer2 to interactively explore an example dataset measuring the gene expression of 2,700 PBMCs obtained from 10X Genomics. Bulk gene expression signatures of cell types from CIBERSORT were used to obtain a tentative cell type for each cell. Please see the video tutorial above for more information.

Clustergrammer2 Roadmap

Please see the project Roadmap for information on where the project is heading.

Installation

A typical installation requires the following commands to be run:

bash
pip install clustergrammer2
jupyter nbextension install --py --sys-prefix clustergrammer2
jupyter nbextension enable --py --sys-prefix clustergrammer2

Optionally: --sys-prefix|--user|--system

Or, if you use jupyterlab:

bash
pip install clustergrammer2
jupyter labextension install @jupyter-widgets/jupyterlab-manager

bash
jupyter nbconvert --to html notebook.ipynb

Development

During development run npm run watch for real time updates. When releasing a new version, first confirm that latest version of the front and back ends are working locally (check console logs, etc).

Updating versions

Update the versions in the following files

JavaScript

package.json
widget.ts

Python

_version.py
example.py
requirements.txt
_frontend.py

Webpack

Run the following commands to build the JavaScript bundle:

bash
npm run build
npm run build:nbextension
npm run build:labextension

Publish to npm using

bash
npm publish

These instructions are based on the release instructions from the jupyter-widgets/widget-ts-cookiecutterREADME.

Bundling the Python Package

Next, bundle the python package using (optinally delete the old versions under dist)

bash
python setup.py sdist bdist_wheel

Then, upload the PYPI:

bash
twine upload dist/*

Checklist after release

After releasing a new version several things need to be checked to ensure proper widget functioning.

Check Package Managers

https://www.npmjs.com/package/clustergrammer2 (can be slow to update)
https://pypi.org/project/clustergrammer2/

Check HTML Embedding

Run jupyter nbconvert --to html introduction_nb.ipynb (from the examples directory) to generate a static HTML and check that the latest widget is working (uses unpkg.com, like https://unpkg.com/clustergrammer2@0.5.18/dist/index.js)
Check NBViewer using link to notebook on GitHub (e.g. https://nbviewer.jupyter.org/github/ismms-himc/clustergrammer2/blob/master/examples/introduction_nb.ipynb?flush_cache=true). Make sure to use the querystring flush_cache=True in the URL to ensure that NBVIewer re-runs nbconvert.