KC Soma Annotations in Neuroglancer • aedes

This vignette identifies soma/root positions for Kenyon cells (KCs) and then adds point annotations to an existing Neuroglancer/Spelunker scene.

The aedes_soma_position() call can use FlyTable soma coordinates, nucleus table, and finally an L2 peoperties + signed neuropil mesh distance fallback for neurons without either of those. For large cell classes, pass a parallel backend via cl and leave the default chunksize = 20L so L2 attributes are fetched in batches.

library(aedes)
library(dplyr)
library(fafbseg)

Find KC Soma Positions

kcs <- aedes_meta("class:KC")
kcsp <- aedes_soma_position(kcs$root_id, cl = 4)
kcsp %>% count(source)

Add Soma Annotations to a Scene

The positions returned by aedes_soma_position() are in nm coordinates by default, so ngl_annotation_layers() should use rawcoords = FALSE.

Use with_aedes() when making annotation layers so that fafbseg uses the Aedes Neuroglancer context, to obtain the correct voxel dimensions.

singletonurl <- "https://spelunker.cave-explorer.org/#!middleauth+https://global.daf-apis.com/nglstate/api/v1/4738104991678464"
rawcoords <- FALSE

kc_ann <- kcsp %>%
  filter(!is.na(position)) %>%
  transmute(
    point = position,
    segments = as.character(root_id),
    layer = paste0("KC soma: ", source)
  )
kc_cols <- setNames(rainbow(n_distinct(kc_ann$layer)), unique(kc_ann$layer))

with_aedes({
  kc_scene <- ngl_decode_scene(singletonurl) +
    ngl_annotation_layers(kc_ann, rawcoords = rawcoords, colpal = kc_cols)
})
kcurl=ngl_encode_url(kc_scene)
browseURL(kcurl)

The resulting scene will have separate annotation layers for each soma source, with source-specific colours:

KC soma annotation layers in Spelunker

If you ask for raw voxel coordinates instead, create kcsp with units = "raw" and set rawcoords <- TRUE before calling ngl_annotation_layers().