We propose to create VOrtex, a Virtual Observatory for the Cortex: Spanning the Scales of Organelles, Cells,
Circuits, and Dynamics. The observatory will disseminate an existing dataset: an automated reconstruction of
all cells in a cubic millimeter of mouse visual cortex, along with the synaptic connectivity of the neurons and
calcium-imaged responses to video stimuli. The cubic millimeter volume spans all layers of cortex and four
visual areas (V1, LM, AL, RL).
A team of human proofreaders will detect and correct the remaining errors in the automated
segmentation. Such semiautomated reconstruction by “proofreading” is much faster than manual tracing of
neurons. Considerable human proofreading effort is still required, however, due to the enormous scale of the
data. Our cubic millimeter volume is 35x larger than an entire Drosophila brain, and 70x larger than the Janelia
fly “hemibrain.”
The proofreading team can be viewed as a new kind of scientific instrument. Just as a telescope is
directed by the astronomy community, proofreading efforts should be directed by the neuroscience community.
Based on requests from users (target cells, scientific rationale), the proofreading schedule will be decided by a
scientific steering committee. The committee will consist of representatives from five subfields: neuronal cell
biology, glial cell biology, cortical cell types, cortical circuits, and theory/simulation. Our novel approach of
centralized labor, but distributed governance, combined with fast public releases of proofreading updates,
should accelerate scientific discovery.
The proofreading team will follow established quality control procedures, and quantitative metrics of
accuracy will be maintained. Updates to the reconstructed wiring diagram will be made on an at least quarterly
basis. Interactive viewing of the EM images and automated segmentation will be possible from any site with an
internet connection. Users may execute large-scale computational analyses on their local clusters or
commercial cloud through programmatic downloading of the data using our CloudVolume software. A
community manager will solicit proofreading requests from the user community, design a proofreading
schedule that maximizes fulfillment of requests, and propose the schedule to the steering committee. The
manager will also recruit users and help them with the mechanics of accessing the data. The Allen Institute will
continue to educate in the use of this data at The Summer Workshop on the Dynamic Brain.
The VOrtex will provide fundamental knowledge about the cortex. The cell biology of neurons and glia
is relevant for brain diseases, and the normal wiring diagram of the cortex could be relevant to the search for
connectopathies.