Figure 2. | 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 2.

Affiliation details

University of California, Berkeley, United States; Northwestern University, United States; Rehabilitation Institute of Chicago, United States
Figure 2.
Download figureOpen in new tabFigure 2. Correct recovery of true numbers of hidden types in synthetic data when incorporating spatial information.

(A) The infinite stochastic block model (which only uses connectivity information) over-estimates the number of classes as it fails to take distance into account, whereas our modeling of the combination of distance and connectivity finds close to the true number of classes. Conn: connectivity; dist: distance. (B) As we increase the true number of types, our method continues to find the correct clustering (as measured by the adjusted Rand index, ARI) whereas the infinite stochastic block model (iSBM) overclusters and thus poorly matches ground truth. (C) We examine the spatial extent (size) of the discovered types (clusters) by measuring the two-dimensional standard deviation of the cell locations. The y-axis indicates what fraction of the discovered types had a given spatial extent. Without incorporating distance, we identify a large number of small, spatially-localized types. With distance, we see a correct recovery of the spatial extent of each type.

DOI: http://dx.doi.org/10.7554/eLife.04250.003