scanpy

What it is and what it does

Scanpy is a Python toolkit for single-cell genomics that handles the full workflow of analyzing gene expression data: reading and preprocessing raw counts, normalizing and scaling, dimensionality reduction, clustering, and statistical testing for differential expression. It is built on top of anndata for efficient data representation and integrates with the broader scverse ecosystem. The package is designed to scale from small pilot studies to datasets with over one million cells, and experimental dask support allows some operations on data larger than available memory.

The toolkit combines visualization (via matplotlib and seaborn), unsupervised learning (via scikit-learn and custom algorithms), and statistical inference (via statsmodels and scipy) into a unified API. It is actively maintained, production-stable, and widely used in academic and research settings for exploratory analysis, cell type discovery, and trajectory inference.

Use it for:

• Preprocessing and quality control of raw single-cell RNA-seq count matrices before downstream analysis.
• Unsupervised clustering and visualization of cell populations to identify cell types or states.
• Differential expression testing between cell groups to find marker genes or disease-associated changes.
• Trajectory inference to reconstruct developmental or differentiation paths from snapshot data.
• Integration with external tools via anndata format for multi-tool pipelines in single-cell studies.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Scanpy is a toolkit for preprocessing, visualizing, clustering, and analyzing single-cell gene expression data, handling datasets from hundreds to millions of cells efficiently.

Yes. Scanpy is a mature, actively maintained toolkit with no known vulnerabilities, permissive licensing, and low install friction. It is the standard choice for single-cell gene expression analysis in Python. Install it if you work with single-cell RNA-seq or similar high-dimensional genomic data; the large dependency tree is a one-time cost for a comprehensive, well-documented analysis platform.

Install

scanpy on PyPI

pip install scanpy

uv add scanpy

poetry add scanpy

Installing scanpy

Before you install

Low install friction with a pure-wheel distribution. Active maintenance with a release 21 days ago and ongoing commits. Requires Python 3.12 or later. Brings 24 runtime dependencies including heavy scientific stacks (numpy, scipy, scikit-learn, pandas, matplotlib), so initial installation is substantial but straightforward.

License in practice

BSD-3-Clause (permissive) allows commercial and private use with minimal restrictions—you must include a copy of the license and the original copyright notice in distributions.

Quickstart

pip install scanpy

import scanpy as sc

adata = sc.read_h5ad('data.h5ad')
sc.pp.normalize_total(adata)
sc.tl.pca(adata)
sc.pl.pca(adata)

Requires Python 3.12 or later. Large dependency tree (numpy, scipy, scikit-learn, pandas, matplotlib, etc.) may take time to resolve and install on first run.

Verify before relying

• Whether dask integration for out-of-core analysis is production-ready or remains experimental as the description suggests.
• Performance characteristics and memory usage on datasets approaching or exceeding one million cells in practice.

Package facts

Evidence: scanpy-1.12.3-py3-none-any.whl

Tags

More Visualization packages