reglscatterpy

Interactive WebGL scatterplots for single-cell / spatial data in Python — AnnData, MuData, SpatialData, pandas and numpy. Renders millions of points in the browser via regl-scatterplot, in Jupyter, JupyterLab, VS Code and Colab.

This is the Python companion to the R package reglScatterplotR — both drive the same compiled widget, so a plot looks and behaves identically across R and Python.

Install

pip install reglscatterpy
pip install anndata          # for AnnData; mudata / spatialdata as needed

30-second start

import scanpy as sc
import reglscatterpy as rs

adata = sc.datasets.pbmc3k_processed()
rs.scatterplot(adata, basis="umap", color="louvain")   # an obs column
rs.scatterplot(adata, basis="umap", color="CST3")      # a gene

Where to next

- :material-book-open-variant: **[User guide](guide.md)** — every feature with runnable examples: big-data rendering, selection round-trip, annotate, DE, linked grids, HTML export, embedding animations. - :material-function: **[Functions API](api.md)** — `scatterplot()` and its 80+ arguments, plus `compose`, `save_html`, `record_html`, `extract`. - :material-cursor-default-click: **[Widget object](widget-api.md)** — the live widget's methods and properties (`selection`, `annotate`, `diff_expression`, `composition`, `highlight`, `morph_to`, `to_html`).

Feature highlights

• Atlas-scale: max_points="auto" density-subsamples while keeping rare cell types; max_points=None puts every cell on the GPU; progressive=True does detail-on-zoom beyond ~4M cells.
• Selection round-trip: lasso in the browser, read w.selection back in Python; annotate, diff_expression, composition on the selection.
• Linked grids: color=[...] or basis=[...] builds a synced multi-panel grid; compose([...]) links arbitrary plots (camera + selection + filters).
• Embedding animation: w.morph_to("spatial") tweens points (and axes) from one embedding to another, preserving filter/selection/colour.
• Offline export: save_html writes a self-contained interactive .html; record_html turns a whole notebook into a static report with no re-run.