Skip to content

Get started

Three pages, in order:

Installation

pip install connecto, plus the optional extras for private annotation sources.

Credentials

connecto reads the CAVE and neuPrint tokens you already have. This page is also where to look when something 401s.

Quickstart

Ten minutes: pick a dataset, select some neurons, pull their connectivity and their morphology.

The idea in one paragraph

Connectomic datasets each speak their own dialect. CAVE (FlyWire, BANC, MICrONS) has materialization versions, root IDs that change under you, and per-datastack table names. neuPrint (hemibrain, maleCNS, MANC, zebrafish) has immutable body IDs, Cypher, and dataset strings like hemibrain:v1.2.1. connecto is a normalisation layer over both: the backends fetch raw frames, and everything you actually touch — column names, dtypes, units, errors — is produced in one place that both backends share.

The practical consequence is that this works:

import connecto as cn

for ds in (cn.FlyWire(), cn.Hemibrain()):
    edges = ds.connectivity.edges("DA1_lPN")
    print(ds.label, edges.shape, list(edges.columns))

FlyWire (FAFB) public release (40581, 3) ['pre', 'post', 'weight']
hemibrain                     (24898, 3) ['pre', 'post', 'weight']

Different backend, different specimen, different ID space — same frame.

What connecto is not

A data-fetching library, and nothing else. No clustering, no matching, no connectivity vectors, no curation policy. That is analysis, and it belongs upstream — see cocoa.