Scanpy

Scanpy is an open-source Python toolkit for the analysis of single-cell transcriptomics data. It is designed to handle large-scale scRNA-seq experiments and is centered on the AnnData data structure, which stores the expression matrix (cells by genes) along with per-cell and per-gene metadata and analysis results.

The library provides an end-to-end, modular workflow for common tasks in single-cell analysis. Typical preprocessing steps include normalization to account for sequencing depth, log transformation, identification of highly variable genes, and scaling. Core analyses cover dimensionality reduction (principal component analysis), construction of a shared nearest-neighbor graph, clustering (notably Leiden and Louvain methods), and visualization using nonlinear embeddings such as UMAP or t-SNE. Scanpy also supports differential expression testing between clusters or conditions and can perform trajectory or pseudotime analysis through diffusion-based methods. For RNA velocity analyses, Scanpy can be used in conjunction with dedicated tools such as scvelo.

Scanpy emphasizes scalability through efficient, sparse-matrix representations and integrates with the broader Python scientific stack, including NumPy, SciPy, scikit-learn, and common plotting libraries. Data are stored in AnnData objects, typically saved in .h5ad files, enabling reproducible and shareable workflows. The project is open-source and widely adopted in academia, with a community of contributors and extensive documentation and tutorials. It interoperates with complementary tools for data integration and batch correction, and is often used alongside other Python-based or R-based single-cell analysis resources to build comprehensive analysis pipelines.