Scanpy

Keep using Scanpy for preprocessing, embeddings and gene ranking. Use GenomeSpy for interactive plots of those results.

Install

pip install --upgrade 'genome-spy-python[scanpy]'

Use that environment as your Jupyter kernel. Scanpy remains an optional dependency.

Replace a plotting call

After computing UMAP with Scanpy, replace sc.pl.umap(...) with:

import genome_spy.integrations.scanpy as gsc

gsc.pl.umap(adata, color="louvain", zoom=True)

The plot reads the existing adata.obsm["X_umap"] coordinates. It does not compute an embedding or change your analysis.

Start with UMAP and embeddings to plot UMAP, t-SNE and diffusion maps, select cells and customize a chart. Use expression plots to compare marker genes across cell groups with dotplots, matrixplots, heatmaps and clustermaps. That guide also covers numeric matrices, correlations, dendrograms and builders. Use expression tracks to inspect individual cells with aligned gene profiles, hover and zoom. Try linked expression summaries to brush a UMAP and compare marker expression within cell types, alongside selected-cell histograms, or highlight the same PBMC3k cells across marker tracks.

Inspect stored PCA diagnostics or plot ranked genes using results already computed by Scanpy.

Explore PAGA graphs and paths, or compare ordered profiles and simulations, including simulation heatmaps and DPT plots.

Use data checks with Scanpy’s full PBMC3k counts. See save and export for interactive HTML, chart definitions and PNG/SVG images, including the plot buttons. Check Scanpy compatibility for available calls and differences from Scanpy. Unsupported options raise an error.