Neuroscience · Interface skeleton
NeuroML
A research platform for connectomics and neural dynamics: mouse-brain records assembled from public datasets, and the published fruit-fly connectome. This page is the layout the research team will implement against. No dataset has been ingested, so no neuron, synapse or model result is displayed.
Development status · layout only. Every panel below states what it will show and which dataset it must come from. Record counts are zero because none have been ingested yet — the zeros are the current true state, not a loading placeholder.
Mouse brain
No single public mouse dataset is a whole-brain synaptic reconstruction, so the platform must keep every source separate, versioned and labelled by what it actually measured. The registry below names the datasets the reconstruction will draw from and the access route each one requires.
| Dataset | What it provides | Access route | Status |
|---|---|---|---|
| MICrONS microns-explorer.org | Mouse visual cortex: electron microscopy, segmentation, synaptic annotations, reconstructed morphology, functional recordings with stimuli. | CAVEclient / CloudVolume; some access needs an authorised CAVE token. | Adapter not implemented |
| Allen Brain Cell Atlas abc_atlas_access | Whole-mouse-brain transcriptomic cell types, spatially resolved molecular data, taxonomies, release labels including beta status. | Public Python package and hosted releases. | Not connected |
| Allen Mouse Brain Connectivity Atlas Allen API | Injection experiments, axonal projections, anatomical targets and region annotations. Projection strength is not a proofread synaptic graph. | Public Allen API. | Not connected |
| Allen Cell Types Database celltypes.brain-map.org | Morphology, electrophysiology, transcriptomic annotation, GLIF and biophysical models, NWB recordings, SWC reconstructions. | Public downloads and API. | Not connected |
| DANDI Archive dandiarchive.org | Standardised NWB neurophysiology recordings with experimental conditions and recording provenance. | Public DANDI API and client. | Not connected |
Coverage
0Neuron records ingested
0Connection records ingested
0 of 5Datasets connected
0Model experiments run
Progress will be reported per dataset and per brain region. There is no defensible whole-brain completion percentage, because the datasets above differ in species preparation, resolution and experimental scope, so no such figure will be shown.
What a neuron record will carry
Identity and provenance
Stable source identifier, internal identifier, dataset and release version, species, brain region, coordinate reference system, source publication, access location, proofreading and reconstruction status.
Biology
Cell type and how it was assigned, dendritic and axonal morphology, synaptic partners with measured synapse counts, electrophysiological properties, transcriptomic annotation, session and subject identifiers where public.
Evidence
An explicit evidence state on every record and every edge, plus a separate layer for predictions. A predicted connection is never written into the verified-connectome layer.
Evidence states
Every visible neuron and connection must say which kind of statement it is. These are the categories the interface will use.
- Verified observationMeasured and recorded in a source dataset, traceable to that record.
- Curated observationReconstructed or annotated by dataset curators from measured material.
- Experimentally supportedBacked by an independent published experiment rather than one measurement.
- Computationally derivedCalculated by our software from measured inputs; reproducible, not observed.
- AI predictedOutput of a trained model, pending experimental support. Always a separate layer.
- Literature inferredDrawn from published findings that may not apply to this animal or preparation.
- Conflicting evidenceSources disagree; both are shown and neither is presented as settled.
- Unknown / unavailableNo data, or data we are not authorised to display. Shown as unavailable, never estimated.
Fruit fly
The published fruit-fly connectome and the earlier fly viewer were built as a separate application. That work is not merged into this site, and this page does not reproduce any of its numbers.
Preserved, not rebuilt
The existing fly implementation stays intact while the platform is expanded. Integration starts with an audit of which dataset release and annotation version it actually loads.
Dataset integrity first
On integration the platform will state the release name, annotation status, neuron and connection counts, and the real graph coverage — quoted from the release, never rounded up.
No fly figures quoted here
No neuron or synapse totals appear on this page because none have been verified against the release that will be used. They appear only when read from a versioned dataset record.
Public API and developer access
The platform is being designed to be usable outside this website: a versioned REST API, documented contracts and explicit status for anything not yet implemented. The endpoint plan and its rules are on the developer contract page. Nothing there responds yet.
Research support
NeuroML is a research programme in development. If you would like to support the work, the funding page is external and contribution there is voluntary: support open research in computational biology and neuroscience ↗. No fundraising target is claimed on this site. The alternative Lightning route is described on research support.
What this page does not claim
- It is not a whole-mouse-brain reconstruction at synaptic resolution. No such public dataset exists.
- It contains no simulated neural activity and no fabricated neurons. Panels are empty because the data is not ingested.
- Predictions, when they exist, will be labelled as predictions and kept apart from measured records.
- Nothing here is a medical device, diagnosis or clinical tool.
Corrections and dataset suggestions: service@orrbiologicals.com. Read the scientific disclaimer and the literature search used for source checking.