Released datasets and visualizations in Neuroglancer

The datasets are available to view interactively using our web-based program Neuroglancer; see GitHub for documentation, or use the upper-right question mark button for quick tips.

Common Layers

  1. 4nm EM: Acquired via multi-beam SEM at a resolution of 4nm*4nm*33nm
    • gs://h01-release/data/20210601/4nm_raw
  2. Masking model: Masking model that identifies neuropil, nuclei, blood vessels, myelin, and fissures in the data.
    • gs://h01-release/data/20210601/masking
  3. 104 proofread cells: 104 manually traced and proofread cells
    • gs://h01-release/data/20210601/proofread_104
    • gs://h01-release/data/20210601/proofread_104/mesh
    • gs://h01-release/data/20210601/proofread_104/skeletons
    • gs://h01-release/data/20210601/proofread_104/subcompartments
  4. Cortical layers: Volumetric boundaries describing white matter and cortical layers 1 through 6
    • gs://h01-release/data/20210601/layers
    • gs://h01-release/data/20210601/layers/mesh
  5. Cell bodies: Locations of 50k cell bodies that reside within the volume (somas only)
    • gs://h01-release/data/20210601/cell_bodies
    • gs://h01-release/data/20210601/cell_bodies/mesh
  6. Cells located within blood vessels: Nuclei annotations for cells identified to be within blood vessels
    • gs://h01-release/data/20210601/blood_vessels
    • gs://h01-release/data/20210601/blood_vessels/mesh
  7. Blood vessel masks: Segmentation of all blood vessels identified in the sample
    • gs://h01-release/data/20210601/blood_vessels_segmented
    • gs://h01-release/data/20210601/blood_vessels_segmented/mesh

Simple view with the “104” library

C2 Sementation Layers

More aggressive agglomeration with a higher likelihood complete automated reconstruction at the expense of additional merge errors.

  1. c2 segmentation: Neuropil segmentation
    • gs://h01-release/data/20210601/c2/
    • gs://h01-release/data/20210601/c2/mesh
  2. skeletons: Skeletonized neuropil
    • gs://h01-release/data/20210601/c2/skeletons
  3. 6-class subcompartments: Volumetric and skeleton labels corresponding to axon, dendrite, soma, astrocyte, axon initial segment, and cilia
    • gs://h01-release/data/20210601/c2/subcompartments
    • gs://h01-release/data/20210601/c2/subcompartments/annotations
  4. Cilia annotations: Point annotations denoting the location of cilia identified by the 6-class subcompartment model
    • gs://h01-release/data/20210601/c2/subcompartments/annotations/cilia
  5. Synapses as points: Point annotations at synaptic locations with associated excitatory or inhibitory type
    • gs://h01-release/data/20210601/c2/synapses/precomputed
  6. Incoming synapses as meshes: Volumetric and meshes of incoming synapses associated with each c2 segment id, separated by excitatory and inhibitory
    • gs://h01-release/data/20210601/c2/synapses/incoming_excitatory
    • gs://h01-release/data/20210601/c2/synapses/incoming_excitatory/meshes
    • gs://h01-release/data/20210601/c2/synapses/incoming_inhibitory
    • gs://h01-release/data/20210601/c2/synapses/incoming_inhibitory/meshes
  7. Volumetric E/I Synapses: Volumetric rendering of pre- and postsynaptic sites
    • gs://h01-release/data/20210601/c2/synapses/whole_ei_onlyvol

c2 library workspace

C3 Segmentation Layers

More conservative agglomeration biased towards fewer false mergers; suggested for connectomic analysis.

  1. c3 segmentation: Neuropil segmentation
    • gs://h01-release/data/20210601/c3/
    • gs://h01-release/data/20210601/c3/mesh
  2. skeletons: Skeletonized neuropil
    • gs://h01-release/data/20210601/c3/skeletons
  3. 6-class subcompartments: Volumetric and skeleton labels corresponding to axon, dendrite, soma, astrocyte, axon initial segment, and cilia
    • gs://h01-release/data/20210601/c3/subcompartments
    • gs://h01-release/data/20210601/c3/subcompartments/annotations
  4. Synapses as points: Point annotations at synaptic locations with associated excitatory or inhibitory type
    • gs://h01-release/data/20210601/c3/synapses/precomputed
  5. Incoming synapses as meshes: Volumetric and meshes of incoming synapses associated with each c2 segment id, separated by excitatory and inhibitory
    • gs://h01-release/data/20210601/c3/synapses/incoming_excitatory
    • gs://h01-release/data/20210601/c3/synapses/incoming_excitatory/meshes
    • gs://h01-release/data/20210601/c3/synapses/incoming_inhibitory
    • gs://h01-release/data/20210601/c3/synapses/incoming_inhibitory/meshes
  6. Volumetric E/I Synapses: Volumetric rendering of pre- and postsynaptic sites
    • gs://h01-release/data/20210601/c3/synapses/whole_ei_onlyvol
  7. Embeddings: Self-supervised SimCLR embeddings for local neuropil regions (explore on the embeddings page)
    • gs://h01-release/data/20210601/c3/embeddings/combined_umap
  8. Synaptic connections database: Export of synaptic connections in Apache Avro format
    • gs://h01-release/data/20210601/c3/synapses/exported/

c3 library workspace

Combined neuroglancer workspace

A default neuroglancer workspace containing all of the above layers is available here.

Data access from Python via TensorStore

We have prepared a Colaboratory notebook that demonstrates how to access the released datasets via TensorStore, a library for reading and writing large multi-dimensional arrays.

Data can also be downloaded directly from Google Cloud Storage (e.g. via gsutil) from the links listed in Neuroglancer, e.g. for the segmentation: gs://h01-release/data/20210601/4nm_raw. The format specification for the data is described here.