Skip to content

Quickstart

Ten minutes, end to end. Every output on this page is copied from a real run.

1. Open a dataset

import connecto as cn

fw = cn.FlyWire()      # neuPrint - FlyWire (FAFB) public release
hb = cn.Hemibrain()    # neuPrint - hemibrain

FlyWire is served by both backends, and neuPrint is the default — it is the same release, and it answers faster. Pass backend="cave" for the chunkedgraph, proofreading and per-synapse transmitters, which the neuPrint mirror does not have. connecto will tell you when you need it.

A dataset is an immutable handle. It holds no neurons, no selection, no cached frame of annotations — just "which connectome, which backend, which version".

fw

<CAVEDataset 'FlyWire (FAFB) public release' backend=cave version=783>

It pinned itself to a version on first access. FlyWire's public release ships materializations 630 and 783; connecto took the newest.

2. Select some neurons

Everything takes the same mini-language, and it always returns IDs:

fw.ids("DA1_lPN")

array([720575940603231916, 720575940604407468, 720575940605102694,
       720575940613345442, 720575940614309535, 720575940617229632,
       720575940619385765, 720575940621185050, 720575940621239679,
       720575940623303108, 720575940623543881, 720575940626034819,
       720575940630066007, 720575940637208718, 720575940637469254])

Fifteen DA1 projection neurons. ids() is a pure function — it returns an array and mutates nothing.

fw.ids("DA1_lPN", side="left")        # 8 of them
fw.ids("/^AOTU00.*")                  # regex -> 35 neurons
fw.ids("super_class:visual_projection")   # any annotation column

3. Ask what they connect to

edges = fw.connectivity.edges("DA1_lPN")
edges.head()

                  pre                post  weight
0  720575940605102694  720575940646122804      64
1  720575940603231916  720575940629163931      50
2  720575940604407468  720575940631143213      49
3  720575940623303108  720575940612355507      48
4  720575940603231916  720575940635945919      46

Three columns — pre, post, weight — as int64, int64, int32. Always. This is a closed schema: FlyWire's underlying edge view carries seventeen extra columns of neurotransmitter probabilities, and connecto drops them so that the frame you get from FlyWire is the frame you get from hemibrain.

It tells you what it dropped, rather than pretending they never existed:

edges.attrs["connecto"]["dropped_columns"]

['gaba', 'ach', 'glut', 'oct', 'ser', 'da', 'connection_score', 'cleft_score',
 'gaba_std', 'ach_std', 'glut_std', 'oct_std', 'ser_std', 'da_std',
 'connection_score_std', 'cleft_score_std', 'valid_nt']

Pass extra=True to keep them.

4. Do it again on a different brain

Same line, different animal:

hb.connectivity.edges("DA1_lPN").head()

          pre        post  weight
0  1765040289  1671620613      84
1   754538881  1671620613      76
2   754534424  1704347707      74
3  1734350908  1671620613      73
4   754534424  1671620613      70

Different backend (neuPrint, not CAVE), different ID space (9-digit body IDs, not 18-digit root IDs), different specimen. Identical frame.

5. Get the morphology

skels = fw.skeletons.get(fw.ids("DA1_lPN")[:2])
skels

<class 'navis.core.neuronlist.NeuronList'> containing 2 neurons (175.8KiB)
               type                name                  id  n_nodes  n_branches  cable_length        units
0  navis.TreeNeuron  720575940603231916  720575940603231916     3588         586    2050971.75  1 nanometer
1  navis.TreeNeuron  720575940604407468  720575940604407468     3909         670    2367593.25  1 nanometer

These are ordinary navis TreeNeurons, in nanometres, so the whole navis ecosystem works with no conversion step:

import navis
navis.plot3d(skels)

Meshes work the same way and give you MeshNeurons.

6. Notice what it refuses to do

mic = cn.MICrONS()
mic.connectivity.synapses(ids, min_score=100)

CapabilityError: MICrONS (minnie65) public does not support `min_score`
(no synapse_scores). Drop the argument, or use a dataset that has it.

MICrONS has no per-synapse cleft scores. A library that quietly ignored min_score here would hand you an unfiltered result while you believed you had filtered it — so connecto raises instead. Defaults never raise; only an explicit request for something the dataset lacks does.

Whole namespaces are absent rather than broken, so you can check first:

hasattr(cn.FlyWire(backend="cave"), "proofreading")   # True  - FlyWire is still edited
hasattr(hb, "proofreading")                           # False - hemibrain is frozen

Where one word covers two promises, connecto splits it. Both datasets have a segmentation volume; only FlyWire has a chunkedgraph under it:

hb.segmentation.locs_to_segments(tbars)   # fine - what body is at this point?
hb.segmentation.update_ids([1734350788])  # CapabilityError: no chunkedgraph

And a capability belongs to a (dataset, backend) pair, not to a dataset — so the same FlyWire has a chunkedgraph through one door and not the other, and the error says which:

cn.FlyWire().segmentation.update_ids(old_ids)

CapabilityError: FlyWire (FAFB) public release (neuprint) does not support
chunkedgraph. Available: annotations, connectivity, meshes, neuroglancer,
roi_connectivity, rois, segmentation, skeletons, somas, synapse_scores, synapses.
The cave backend does: cn.get_dataset("flywire", backend="cave").

Where to go next