Scanpy – Single-Cell Analysis in Python¶

Scanpy is a scalable toolkit for analyzing single-cell gene expression data built jointly with anndata. It includes preprocessing, visualization, clustering, trajectory inference and differential expression testing. The Python-based implementation efficiently deals with datasets of more than one million cells.

Discuss usage on Discourse and development on GitHub.
Get started by browsing tutorials, usage principles or the main API.
If Scanpy is useful for your research, consider citing Genome Biology (2018).
Follow changes in the release notes.
Find tools that harmonize well with anndata & Scanpy via the external API and the ecosystem page.

News - CZI grant 2019-11-14¶

Scanpy has been selected an essential open source software for science by CZI among 32 projects, along with giants such as Scipy, Numpy, Pandas, Matplotlib, scikit-learn, scikit-image/plotly, pip, jupyterhub/binder, Bioconda, Seurat, Bioconductor, and others.

Latest additions¶

On master¶

Bug fixes

Bumped version requirement of scipy to scipy>1.4 to support rmatmat argument of LinearOperator issue 1246

1.5.1 2020-05-21¶

Bug fixes

Fixed a bug in pca(), where random_state did not have an effect for sparse input PR 1240 I Virshup
Fixed docstring in pca() which included an unused argument PR 1240 I Virshup

1.5.0 2020-05-15¶

The 1.5.0 release adds a lot of new functionality, much of which takes advantage of anndata updates 0.7.0 - 0.7.2. Highlights of this release include support for spatial data, dedicated handling of graphs in AnnData, sparse PCA, an interface with scvi, and others.

Spatial data support

Basic analysis → tutorial: spatial/basic-analysis and integration with single cell data → tutorial: spatial/integration-scanorama G Palla
read_visium() read 10x Visium data PR 1034 G Palla, P Angerer, I Virshup
visium_sge() load Visium data directly from 10x Genomics PR 1013 M Mirkazemi, G Palla, P Angerer
spatial() plot spatial data PR 1012 G Palla, P Angerer

New functionality

Many functions, like neighbors() and umap(), now store cell-by-cell graphs in obsp PR 1118 S Rybakov
scale() and log1p() can be used on any element in layers or obsm PR 1173 I Virshup

External tools

scvi() for preprocessing with scVI PR 1085 G Xing
Guide for using Scanpy in R PR 1186 L Zappia

Performance

pca() now uses efficient implicit centering for sparse matrices. This can lead to signifigantly improved performance for large datasets PR 1066 A Tarashansky
score_genes() now has an efficient implementation for sparse matrices with missing values PR 1196 redst4r.

Warning

The new pca() implementation can result in slightly different results for sparse matrices. See the pr (PR 1066) and documentation for more info.

Code design

stacked_violin() can now be used as a subplot PR 1084 P Angerer
score_genes() has improved logging PR 1119 G Eraslan
scale() now saves mean and standard deviation in the var PR 1173 A Wolf
harmony_timeseries() PR 1091 A Mousa

Bug fixes

combat() now works when obs_names aren’t unique. PR 1215 I Virshup
scale() can now be used on dense arrays without centering PR 1160 simonwm
regress_out() now works when some features are constant PR 1194 simonwm
normalize_total() errored if the passed object was a view PR 1200 I Virshup
neighbors() sometimes ignored the n_pcs param PR 1124 V Bergen
ebi_expression_atlas() which contained some out-of-date URLs PR 1102 I Virshup
ingest() for UMAP 0.4 PR 1165 S Rybakov
louvain() for Louvain 0.6 PR 1197 I Virshup
highly_variable_genes() which could lead to incorrect results when the batch_key argument was used PR 1180 G Eraslan
ingest() where an inconsistent number of neighbors was used PR 1111 S Rybakov