NGSidekick

Tools for working with neuroglancer. See docs.

Feature Highlights

• Segment properties: read and write neuroglancer's precomputed segment properties format from a pandas DataFrame.

• Local annotations: construct local annotations directly in viewer state. (API subject to change.)

• Precomputed annotations: export annotations in neuroglancer's precomputed annotations format from a pandas DataFrame.

  • Supports all five annotation types:
    • point
    • line
    • axis_aligned_bounding_box
    • ellipsoid
    • polyline
  • Per-annotation properties (numeric, enum/categorical, and rgb/rgba color).
  • Per-annotation relationships (lists of related segment IDs, used by neuroglancer to filter annotations by segment).
  • Written to all "index" types (annotation id, related segment, and multi-level spatial grid)
  • Sharded output, written in parallel via tensorstore.
  • Note: writing directly to cloud storage is not yet supported; outputs must be written to a local filesystem (you must upload to cloud storage afterwards).

Installation

Packages are available from both PyPI and conda-forge.

Using pixi:

pixi add ngsidekick

Using conda:

conda install -c conda-forge ngsidekick

Using pip:

pip install ngsidekick

Using uv:

uv add ngsidekick

# or in an existing environment
uv pip install ngsidekick