Summary of connectivity and size by depth (related to Fig. 5, Supplementary Fig. 1)

To produce the synapse distribution by depth, featured prominently in the Cell Type Catalog (Supplementary Fig. 1), we first identify all synapses of each target cell from the designated cell type. Using a k-dimensional tree, we find the nearest neighboring column pin by Euclidean distance for each presynapse and postsynapse. The depth property of the column pin associates each synapse with one of the 121 medulla, 76 lobula, or 51 lobula plate depth bins (see Defining column centerlines and ROIs). Across all neurons of a cell type, we calculate the mean count separately for presynapses and postsynapses, per depth. We smooth the distribution per brain region with a Savitzky-Golay filter (window size 5, 1st-order polynomials) before plotting on the left-side panel of the summary data figures. Separately we provide the percentage of synapses located within the AME over the cell type’s total number of synapses, again separated by presynapses and postsynapses.

To quantify each cell type’s size as a function of depth, we find the depth for all synapses. Since columns are implemented as ROIs in neuPrint (see Connectome data access overview), each synapse comes with an associated column. For each cell type, we calculate the mean count of columns with synapses (combining pre and post) per depth and smooth the distribution per brain region with a Savitzky-Golay filter (window size 5, 1st-order polynomials) before plotting on the right-most panel of the summary data.

The center panel of the cell type summary contains a connectivity summary. For all neurons of a target cell instance (effectively type, but separately accounting for right and left instances), we find all synapses and their connecting partner cells. We order them by the fraction of input and output connections and show the top five after removing unnamed segments.