Build a BANC Neuroglancer scene with a custom LM image layer

Construct a Neuroglancer scene that overlays a light-microscopy precomputed image layer (e.g. the output of neuronbridger::nrrd_to_precomputed applied to a registered confocal stack) on top of the canonical public BANC scene from banc_scene (BANC EM, segmentation proofreading, region outlines, JRC2018F atlas, FAFB / hemibrain / MANC imported meshes, synapse cloud, nuclei) and either return a long fragment-encoded URL or, when shorten = TRUE, POST the state to the BANC state server and return a shortened spelunker.cave-explorer.org/#!middleauth+...nglstate/api/v1/<id> URL — the same URL form bancsee produces.

Usage

banc_lm_scene(
  lm_url,
  layer_name = "LM data",
  shader = NULL,
  range = c(1, 30),
  opacity = 0.55,
  blend = c("additive", "default"),
  volume_rendering = c("on", "max", "off"),
  volume_rendering_depth_samples = 788L,
  volume_rendering_gain = NULL,
  ids = NULL,
  url = NULL,
  shorten = TRUE,
  open = FALSE
)

Arguments

lm_url: URL of the precomputed LM layer (gs://..., https://..., or file:///... for local viewing). precomputed:// is added automatically if missing.
layer_name: display name of the LM layer in Neuroglancer. Default "LM data".
shader: optional Neuroglancer shader string. If NULL (default) the layer uses Neuroglancer's built-in emitGrayscale shader, controlled by shaderControls.normalized.range (set via range).
range: numeric length-2 vector c(low, high) setting shaderControls.normalized.range — the source-data intensity window mapped to 0..1 brightness. Default c(1, 30), which suits LM volumes that have been Elastix-warped + clipped to BANC voxel space (B-spline ringing leaves most of the signal in the bottom tenth of the uint8 range). Increase the upper bound for brighter source data; tighten it for sparse stains.
opacity: layer opacity in [0, 1]; default 0.55 (matching the public BANC JRC2018F atlas imported layer).
blend: layer blend mode ("default", "additive"). Default "additive" so LM signal lights up where it overlaps the EM rather than occluding it.
volume_rendering: one of "on" (default), "max" or "off". Required for the layer to be visible in 3-D Neuroglancer views. Cross-section / orthogonal slice views ignore this setting.
volume_rendering_depth_samples: integer; how many depth samples Neuroglancer uses when ray-tracing the volume in 3-D. Default 788 (the value the public BANC atlas uses).
volume_rendering_gain: numeric; 3-D volume rendering gain. NULL (default) leaves it unset, which lets Neuroglancer's UI control it interactively.
ids: optional vector of BANC root IDs to add to the "segmentation proofreading" layer that ships in the base scene.
url: Spelunker base URL with a starting state to extend. Defaults to the public BANC scene used by banc_scene.
shorten: logical; if TRUE (default), POST the state via the internal banc_shorturl() and return a shortened URL. If FALSE, return the long fragment-encoded URL with the state JSON inlined.
open: logical; if TRUE, open the result in the system browser via browseURL.

Value

A character of length 1: a shortened spelunker.cave-explorer.org/#!middleauth+...nglstate/api/v1/<id> URL by default, or a long fragment URL when shorten = FALSE.

Viewers (`ng.banc.community/view/` / `ng.banc.community/`)

The BANC team also serves a public viewer at ng.banc.community/view/ and a private one at ng.banc.community/. These viewers do not POST states dynamically — they load named states from ngstate.banc.community/view/<state-name>, which redirects to static JSON files committed under the-BANC-fly-connectome/neuroglancer_states/view/. To publish a state to ng.banc.community/view/, save the state JSON returned by this function (set shorten = FALSE and decode the fragment, or use fafbseg::ngl_decode_scene) and PR it into the BANC repo. For ad-hoc / dev sharing, the shorten = TRUE URL is the practical pattern.

Bucket policy

Lee-lab maintains a curated mirror of registered LM volumes at gs://lee-lab_brain-and-nerve-cord-fly-connectome/light_level/, but that bucket is not public-write. To produce a sharable URL of your own you'll need either (a) write access from the lee-lab maintainers, or (b) your own public-read GCS / S3 / static HTTP host. Once the precomputed directory is reachable, pass its URL as lm_url.

Examples

if (FALSE) { # \dontrun{
# Convert + serve a registered LM volume (the upstream
# vignettes do this in detail):
neuronbridger::nrrd_to_precomputed(
  "CapaR_in_JRC2018U_HR.nrrd",
  output     = "/tmp/CapaR_pc",
  resolution = c(519, 519, 1000)   # JRC2018U_HR voxel size in nm
)
system("gsutil -m cp -r /tmp/CapaR_pc gs://your-bucket/lm/CapaR/")

u <- banc_lm_scene(
  "gs://your-bucket/lm/CapaR",
  layer_name = "Kondo 2020 - CapaR",
  open       = TRUE
)
} # }