Quickstart

Every scanpy analysis works on an AnnData object: scanpy.pp preprocesses it, scanpy.tl computes on it and scanpy.pl plots it. The example below clusters 3k PBMCs and finds the marker genes of each cluster.

import scanpy as sc

adata = sc.datasets.pbmc3k()

sc.pp.filter_cells(adata, min_genes=100)
sc.pp.filter_genes(adata, min_cells=3)
sc.pp.normalize_total(adata)
sc.pp.log1p(adata)
sc.pp.highly_variable_genes(adata, n_top_genes=2000)

sc.tl.pca(adata)
sc.pp.neighbors(adata)
sc.tl.umap(adata)
sc.tl.leiden(adata, flavor="igraph", n_iterations=2)
sc.tl.rank_genes_groups(adata, "leiden")

sc.pl.umap(adata, color="leiden")
sc.pl.rank_genes_groups_dotplot(adata, n_genes=3)

Most functions store their results in adata instead of returning them, e.g. the clusters in adata.obs["leiden"]. Leiden clustering needs igraph, which pip install 'scanpy[leiden]' installs, see Installation.

Preprocessing and clustering walks through the same steps with quality control and doublet detection, Usage Principles explains the conventions behind the API and API lists every function.