117 (Jun 2021)

Data Release v117 (June 11, 2021) is the first public release of the dataset.

This version introduces the following CAVE tables:

This data release coincides with the release of many of the Static Data Repositories, including:

Information about the version 117 release can be queried with the following:

# report timestamp metadata on the version of interest
client.materialize.get_version_metadata(117)['time_stamp']
datetime.datetime(2021, 6, 11, 8, 10, 0, 215114, tzinfo=datetime.timezone.utc)

Synapse Table

Table name: synapses_pni_2

The only synapse table is synapses_pni_2. This is by far the largest table in the dataset with 337 million entries, one for each synapse. It contains the following columns:

Synapse table column definitions
Column Description
pre_pt_position / pre_pt_supervoxel_id / pre_pt_root_id The bound spatial point data for the presynaptic side of the synapse.
post_pt_position / post_pt_supervoxel_id / post_pt_root_id IThe bound spatial point data for the postsynaptic side of the synapse.
size The size of the synapse in voxels. This correlates well, but not perfectly, with the surface area of synapse.
ctr_pt_position A position in the center of the detected synaptic junction. Of all points in the synapse table, this is usually the closest point to the surface (and thus mesh) of both neurons. Because it is at the edge of cells, it is not associated with a root id. 
# Synapse query: outputs
client.materialize.synapse_query(pre_ids=example_root_id)

# Synapse query: inputs
client.materialize.synapse_query(post_ids=example_root_id)

For more on how to interpret the table, see Annotation Tables.

Nucleus Table

Table name: nucleus_detection_v0

Nucleus detection has been used to define unique cells in the dataset. Distinct from the neuronal segmentation, a convolutional neural network was trained to segment nuclei. Each nucleus detection was given a unique ID, and the centroid of the nucleus was recorded as well as its volume. Many other tables in the dataset are reference tables on nucleus_detection_v0, meaning they are linked by the same annotation id. The id of the segmented nucelus, a 6-digit integer, is static across data versions and for this reason is the preferred method to identify the same ‘cell’ across time.

The key columns of nucleus_detection_v0 are:

Nucleus table column definitions
Column Description
id 6-digit number of the segmentation for that nucleus; ‘nucleus ID’
pt_position  pt_supervoxel_id  pt_root_id Bound spatial point columns associated with the centroid of the nucleus

Note that the id column is the nucleus ID, also called the ‘soma ID’ or the ‘cell ID’.

# Standard query
client.materialize.query_table('nucleus_detection_v0')

# Content-aware query
client.materialize.tables.nucleus_detection_v0(id=example_nucleus_id).query()

For more on how to interpret the table, see Annotation Tables.

Neuron Nucleus Table

Table name: nucleus_neuron_svm

Nucleus detection classified as likely neuron.

Deprecated table

This table remains available from materialization versions: 117, 343, 795, 943, 1078

Use nucleus_ref_neuron_svm instead, which is a reference on the nucleus table nucleus_detection_v0.

Automated cell-type classifier (coarse, version 1)

Table name: allen_soma_coarse_cell_class_model_v1

This is a model developed by Leila Elabbady and Forrest Collman, it uses features extracted from the somatic region and nucleus segmentation (developed in collaboration with Shang Mu and Gayathri Mahalingam). Those features included the number of soma synapses, the somatic area, the somatic area to volume ratio, the density of somatic synapses, teh volume of the soma, the depth in cortex of the cell (based upon the y coordinate after a 5 degree rotation), the nucleus area, the ratio of the nucleus area to nucleus volume, the average diameter of the proximal processes of the cell, the area of the nucleus with a fold, the fraction of the nucleus area within a fold, the volume of the nucleus, and the ratio of the volume of the nucleus to the volume of the soma. The model was trained using labels from the allen_v1_column_types_v2 table supplemented with NP labels from allen_minnie_extra_types as of version 91 . The model is a SVM classifier, using an rbf kernel, with class balance. On 20% of the data held out the model has a 77% accuracy, with principal confusion between layer 6 IT and CT, and layer 5 IT with layer 4 and many types.

Deprecated table

This table remains available from materialization versions: 117

Use the current version aibs_metamodel_celltypes_v661 instead.

Flat Segmentation

minnie65 flat segmentation v117
Name Volume Cloudpath Short Description Type (size)
Proofread Segmentation (v117) minnie65 https://storage.googleapis.com/iarpa_microns/minnie/minnie65/seg Mulit-resolution flat / static cellular segmentation voxels and meshes from 8,8,40 nm and above Precomputed Shareded Compressed Segmentation (12 TB)

This contains the fixed state of the cellular segmentation at each version, where each voxel has been assigned an ID which is unique to each cellular object at 8,8,40, along with downsampled versions. Not all objects have been proofread, but a summary of the most focused efforts on cells can be found in the proofreading status metadata. In addition the mesh folder contains meshes of each object available at 3 different levels of downsampling. Folder contains many files, for download use cloud-volume, tensor-store, for bulk download use igneous, AWS CLI or gsutil CLI.

Back to top