Clustergrammer2 is a Jupyter Widget that enables researchers to interactively explore and analyze high-dimensional single-cell data. Our goals are to improve and enhance:
- the interactive visualization experience
- the data analysis experience
- the usage, sharing, and deployment experience
Documentation for the Clustergrammer projects can be found at https://clustergrammer.readthedocs.io. Please send feedback or concerns via the Gitter chat room and or the issue tracker.
Last updated: February 4th, 2020
Clustergrammer2 allows users to interactively explore single-cell datasets consisting of millions of data points (e.g. thousands of single cells in thousands of dimensions). Clustergrammer2 uses the JavaScript library Clustergrammer-GL to generate the interactive heatmaps. Clustergrammer-GL is built using the WebGL library regl. Development of the interactive WebGL visualizations are primarily done on the Clustergrammer-GL repo.
We are working on additing additional data analysis methods to facilitate exploration and analysis of single cell data. The Python back-end of Clustergrammer2 handles data analysis tasks (e.g. filtering, normalization, clustering) and development of additional data analysis methods is done in this repo.
We are also working on integration of location-based data (e.g. multiplex ion-beam imaging, CODEX, 10X Genomics Visium) into Clustergrammer's workflows. Location based data holds tremendous promise for investigating cell-to-cell communication (e.g. ligand-receptor interaction inference, see Single-Cell-Immune-Profiling-of-Atherosclerotic-Plaques).
We are working on using the librray voila for generating dashboards. The Jupyter Widgets framework enables communication between widgets and we are currently working on using this feature to build dashboards using Clustergrammer2.
We are currently working on examples Jupyter notebooks using Clustergrammer2 (see Clustergrammer2-Notebooks and Cast Studies and Tutorials). Users can run these example noteooks by installing Clustergrammer2 locally or for free using several cloud-based Jupyter instances:
- MyBinder
- Saturn Cloud
- Kaggle
- Improve API
- testing and continuous integration
- Clear Old WebGL State
- Dendrogram Slicing
- Initial Ordering
The broader Clustergrammer project documentation can be found here clustergrammer.readthedocs.io/.