Installation

To install CellMapFlow, you can use pip:

pip install cellmap-flow

Note that the basic installation does not include DaCapo and BioImage.io core dependencies. To install CellMapFlow with DaCapo support, use the following command:

pip install cellmap-flow[dacapo]

To install CellMapFlow with BioImage.io support, use the following command:

pip install cellmap-flow[bioimageio]

To install CellMapFlow with both DaCapo and BioImage.io support, use the following command:

pip install cellmap-flow[dacapo,bioimageio]

Deployment with pixi

The repository carries a pixi.toml and lockfile describing the environment cellmap-flow is deployed from, including what PyPI cannot provide: the CUDA toolkit, and the MinIO server and client that annotation painting needs. From a checkout:

pixi install          # the default environment: dashboard, catalog and cellpose models, finetuning
pixi run cellmap_flow view -d /path/to/dataset.zarr

runnables.yaml registers the same commands as Fileglancer apps. Inside a pixi environment, inference servers submitted to the cluster start with pixi run cellmap_flow serve (CELLMAP_FLOW_SERVER_COMMAND), so they run from the same lockfile. Command line lists the commands. Models whose packages the default environment lacks (fly_organelles checkpoints, BioImage Model Zoo models, Cellpose 4) are served from the lockfile’s other environments; see Running a model in its own environment.