Figure 1. | Automatic discovery of cell types and microcircuitry from neural connectomics

Open accessCopyright infoDownload PDF

Automatic discovery of cell types and microcircuitry from neural connectomics

Figure 1.

Affiliation details

University of California, Berkeley, United States; Northwestern University, United States; Rehabilitation Institute of Chicago, United States
Figure 1.
Download figureOpen in new tabFigure 1. Deriving circuitry and cell types from connectomics data.

(A) As input we take the connectivity between cells (B), the distance between them (C), the depth of the cell bodies (D), and the depth profile of the synapses (E). (F) Our algorithm discovers hidden cell types in this connectivity data by assuming all cells of a type share a distance-dependent connectivity profile, similar depth, and a similar synaptic density profile, with cells of other types. This results in a clustering of the cells by those hidden types. (F) Shows the cell connectivity matrix with cells of the same type grouped together. (G) Shows the learned probability of connection (p(conn)) between our different types at various distances—in this case, the cells are likely to connect when they are close. (H) Shows the probability of connection (p(conn)) between two cell types that very rarely connect—there is a background ‘base’ connection rate to account for errors in data, but the probability is very low. (I) Shows that we also recover the expected laminarity of types and the depth-specific (J) synaptic connectivity. (K) We then plot how the connectivity between these types changes as a function of distance between the cell bodies to better understand short-range and long-range connectivity patterns.