Data & provenance
This application draws on third-party scientific datasets. It does not own them, and it does not redistribute them. Everything below states what each source actually contains, what this deployment can currently reach, and how a figure on screen should be read.
Provenance classes
Every quantity shown anywhere in the interface carries one of these labels. They are ordered from strongest to weakest claim.
| Class | Meaning |
|---|---|
| MEASURED | Directly observed in the source dataset. |
| DERIVED | Computed deterministically from measured data. |
| PREDICTED | Output of a forecasting model, not an observation. |
| SIMULATED | Output of a computational model of the connectome. |
| INFERRED | Our interpretation of a pattern. Not a measurement, and not a claim about what the animal experiences. |
Data origin badges
| Badge | Meaning |
|---|---|
| LIVE | Queried directly from the upstream connectome service. |
| EXPORT | Pipeline export of the published dataset. |
| DEVELOPMENT SAMPLE | Procedurally generated placeholder with the same schema as the real dataset. Contains no biological measurements. |
| SYNTHETIC BENCHMARK | Deterministic synthetic population generated to measure rendering performance. Contains no biological measurements. |
Fish1 — status in this deployment
| Datastack | fish1_v250915 |
| Materialization | unavailable |
| Connectivity | unavailable |
| Skeletons | unavailable |
| Whole-brain index | not exported — run pipeline/fish1/export_neurons.py |
| Voxel size | 8 × 8 × 30 nm |
| Axis orientation | not documented by the release; no anatomical direction labels are shown |
Fish1 — what it contains
A correlated light and electron microscopy dataset from a 7 dpf larval zebrafish, covering the brain and anterior spinal cord. It is structural only: it records anatomy and synaptic connectivity, and contains no neural activity, no behaviour, and nothing resembling an internal state.
| Table | Contents |
|---|---|
| somas | One row per segmented soma: stable lore ID, cell type (exc / inh / na), current root ID, supervoxel, voxel position. |
| synapses_axde_label | Axon-to-dendrite synapses with pre/post root IDs, positions, and an excitatory/inhibitory tag. |
| synapses_axon_to_dendrite_size | Per-synapse bounding boxes, used to filter out small or spurious contacts. |
Molecular identity comes from genetic labelling: vglut2a+ for excitatory, gad1b+ for inhibitory. A cell typed na was simply not annotated — that is not a claim that it is neither.
Licence: open access. Citation is required.
Petkova, M. D., Januszewski, M., et al. (2025). A connectomic resource for neural cataloguing and circuit dissection of the larval zebrafish brain. bioRxiv. paper · release
Identifiers
| Identifier | Behaviour |
|---|---|
| lore ID | Small stable integer naming a soma. Unchanged by proofreading. This is what the application persists and what appears in shareable URLs. |
| root ID | 64-bit ID of the current segmentation of a neuron. Changes after merge or split edits, and is only meaningful together with a materialization version. |
| supervoxel ID | Leaf chunk of the segmentation graph. |
Other datasets
Whole-brain calcium imaging of over 70,000 neurons under visual stimuli. Functional data, published as a forecasting benchmark. Not ingested in this build.
A different animal from Fish1. There is no cell-level correspondence between the two, and this application will not paint ZAPBench traces onto Fish1 soma.
Connectivity and activity from the same individual animal, aligned per cell. This is the dataset that would make a genuine stimulus-to-behaviour story possible.
Access is restricted. This application holds none of it, downloads none of it, and contains no endpoints for it. A typed adapter boundary exists and nothing more.
A deterministic synthetic population with the same schema as a real export, shaped like a larval zebrafish brain so that rendering and culling behave realistically.
Contains no biological measurements. Always badged in the viewport.
What this application does not claim
No dataset here reports what a fish feels, wants, or is thinking about. Terms such as escape response or locomotor drive describe circuit and behavioural states, and when they eventually appear they will be labelled INFERRED, carry the evidence they were computed from, and never be presented as a measurement.
Graph distance is not time. If a future view animates signal propagation by traversal depth, that is a depth counter, not milliseconds of conduction, and it will be labelled as such unless an explicit model produces the timing.