scanpy

Why we included this project

Scanpy is the closest thing single-cell genomics has to a standard toolchain, and much of the field's published work builds on it. It bundles preprocessing, visualization, clustering, trajectory inference, and differential expression testing into one Python library, so you can move from raw counts to interpretable results without wiring together a handful of packages. Everything is built around the AnnData object, which keeps expression matrices together with their per-cell and per-gene annotations and makes it easy to pass data between Scanpy and the rest of the scverse ecosystem. It also holds up on genuinely large datasets: it handles over a million cells comfortably, and an experimental Dask backend extends many functions to data that will not fit in memory. Because it is so widely taught and used, pipelines you write here are easy to share and compare against published analyses.