Installation¶

connecto needs Python 3.11 or newer.

pipuvFrom source

pip install connecto

uv add connecto

git clone https://github.com/schlegelp/connecto
cd connecto
pip install -e ".[dev]"

What comes with it¶

These are hard dependencies, not extras:

package	why
`caveclient`	the CAVE backend — FlyWire, BANC, MICrONS
`neuprint-python`	the neuPrint backend — hemibrain, maleCNS, MANC, fish2
`navis`	skeletons and meshes come back as navis neurons
`DracoPy`	meshes are draco-encoded everywhere connecto reads them
`pandas`, `numpy`, `pyarrow`, `networkx`, `trimesh`, `tqdm`	the usual

connecto does not use cloud-volume

Segmentation volumes and meshes are read by connecto.precomputed, connecto's own reader for the neuroglancer precomputed and graphene formats. cloud-volume also writes, to Google Cloud, S3 and half a dozen other backends, and pays for that in dependencies — boto3, the Google Cloud SDK, gevent, protobuf and the rest come to roughly 78 MB across some 37 packages, none of which reading needs.

connecto's reader is checked against cloud-volume rather than instead of it: tests/test_precomputed.py reads the same cutouts and the same meshes both ways and asserts they agree, so cloud-volume is a development dependency.

caveclient must be ≥ 8.0

This is pinned for a reason. Before 8.0, caveclient streamed query results as CSV and re-inferred the types on the way back in — which silently truncated int64 root IDs. 8.0 switched to a pandas-native path. If you have an older caveclient pinned by something else in your environment, connecto's root IDs can come back subtly wrong rather than obviously broken.

Optional extras¶

Some annotation sources are lab-internal or need an extra package:

pip install "connecto[voxels]"    # compiled decoders for dense reads
pip install "connecto[clio]"      # Clio annotations (maleCNS)
pip install "connecto[flytable]"  # SeaTable / "flytable" annotations
pip install "connecto[points]"    # parallel point -> segment lookups

Nothing in the core depends on these. If you ask for a source you have not installed, you get a MissingDependencyError telling you the exact pip install to run — not an ImportError from three frames down.

Check it worked¶

import connecto as cn

cn.list_datasets()

                 name                          label                  species        backends                     annotations  public
0               aedes         Aedes (mosquito brain)            Aedes aegypti            cave                        flytable   False
1                banc    BANC (brain and nerve cord)  Drosophila melanogaster  neuprint, cave                  cave, flytable    True
2                fanc              FANC (female VNC)  Drosophila melanogaster            cave                            cave   False
3               fish2       fish2 (larval zebrafish)              Danio rerio        neuprint                        neuprint   False
4             flywire  FlyWire (FAFB) public release  Drosophila melanogaster  neuprint, cave                public, flytable    True
5  flywire-production      FlyWire (FAFB) production  Drosophila melanogaster            cave                public, flytable   False
6           hemibrain                      hemibrain  Drosophila melanogaster        neuprint                        neuprint    True
7             malecns                       male CNS  Drosophila melanogaster        neuprint                  neuprint, clio    True
8                manc                MANC (male VNC)  Drosophila melanogaster        neuprint                        neuprint    True
9             microns      MICrONS (minnie65) public             Mus musculus            cave  celltypes, mtypes, nucleus_svm    True

The first backend listed is the default. public=False does not mean secret — it means a fresh token will not get you in, and cn.get_spec(name).access says what will.

That call is offline — it only reads the registry. To actually query anything you need tokens, which is the next page.