Get started¶
Four pages, in order. The first three are enough to open a dataset and look at it; the last only matters once you want to write annotations back somewhere.
In a hurry?
uvx bigclust2@latest --from https://flyem.mrc-lmb.cam.ac.uk/flyconnectome/bigclust_data/examples/MaleCNS_FlyWire_hemibrain_central_brain_bigclust
That opens a public example dataset — 87,000 central brain neurons co-clustered across three connectomes. No account, no credentials, nothing to build.
Installation¶
uvx bigclust2@latest and you are running. What uvx actually does, how to pin
a version, and how to run from a clone.
The example dataset¶
A public co-clustering of FlyWire, MaleCNS and Hemibrain neurons you can open with one command. The fastest way to see what the app does.
Your first project¶
Open a dataset, find your way around the window, select some points and see the neurons behind them.
Credentials¶
Tokens for neuPrint, FlyTable and Clio — needed for refreshing meta data and for pushing annotations, and for nothing else.
What you need before you start¶
Something to open. BigClust opens a directory — local or served over
HTTP — containing an info file and at minimum a meta table. It does not open
a bare .parquet or a distance matrix you hand it.
You do not need one of your own to start: the example dataset is a public URL and needs nothing but the command above. When you are ready to build your own, create a local dataset walks through a complete worked example, the data format reference is the specification behind it, and why data sources are directories explains the reasoning.
Python 3.10 or newer, though if you install with uvx you never have to
think about it — uv fetches an interpreter itself.
A GPU that can run WebGPU. The scatter plot and the 3D viewer are both rendered with pygfx on wgpu. Any machine from the last decade is fine; a headless server without a display is not.
Nothing else. No account, no configuration file, no credentials — those only enter the picture at the point where BigClust needs to talk to an annotation backend, which is refreshing meta data and pushing annotations.
Next: installation.