Installation

You need Python 3.13 or newer. If you don’t have Python installed, we recommend uv. PyPI is the supported install. Cloning the repository and running uv sync is for contributors, see Contributing.

Alpha release

mantispy is in alpha, and its releases are pre-releases (0.1.0aN). While no final release exists, pip and uv both fall back to the newest pre-release, so a plain install works:

uv add mantispy  # in a uv project; `uv init` creates one
pip install mantispy  # or with pip

Pass --pre (uv pip install --pre mantispy, or uv add --prerelease allow mantispy in a project) to get pre-releases of mantispy’s dependencies as well, or to keep getting the newest alpha once a final release is out.

Development version

Install the latest development version from GitHub:

uv add git+https://github.com/scverse/mantispy  # in a uv project
pip install git+https://github.com/scverse/mantispy.git  # or with pip

Optional dependencies

Images and segmentations

read_plate(), jump_plate() and blobs() read images into SpatialData:

uv add 'mantispy[spatial]'  # in a uv project
pip install 'mantispy[spatial]'  # or with pip

Mean average precision

map() runs copairs, an optional extra:

uv add 'mantispy[map]'  # in a uv project
pip install 'mantispy[map]'  # or with pip

Integration metrics

evaluate_integration() delegates to scib-metrics:

uv add 'mantispy[integration]'  # in a uv project
pip install 'mantispy[integration]'  # or with pip

Interactive plots

With plotly installed, plots in a Jupyter notebook also draw an interactive version:

uv add 'mantispy[interactive]'  # in a uv project
pip install 'mantispy[interactive]'  # or with pip

Tutorials

The single-cell tutorial clusters cells with scanpy.tl.leiden(..., flavor="igraph"), which needs igraph:

uv add igraph  # in a uv project
pip install igraph  # or with pip

To run every tutorial and case study, install all of the above at once:

uv add 'mantispy[map,integration,interactive]' igraph  # in a uv project
pip install 'mantispy[map,integration,interactive]' igraph  # or with pip