FAFBseg tools#

fafbseg is a set of Python tools to work with various kinds of segmentation data in the FAFB dataset:

  1. Google’s auto-segmentation

  2. FlyWire by the Seung/Murthy labs

  3. Buhmann et al. (2020) synapse predictions

A lot of this library depends on services hosted by Eric Perlman and Davi Bock. So if you happen to bump into them, buy them a drink ;)

Check out the introduction for a brief overview of FAFB data and the tutorials for code examples.

Features#

  • map locations to Google segmentation / FlyWire root IDs

  • transform neurons between FAFB and FlyWire space

  • merge neurons from Google autoseg into v14 main CATMAID instance

  • load FlyWire neurons and skeletonize them

  • generate connectivity tables using the Buhmann et al. synapse predictions

  • parse and generate FlyWire URLs

fafbseg is based on navis and produces objects (skeletons, meshes, etc.) that can be directly plugged into navis’ functions. This in turn enables you among other things to:

  • run morphological clustering/matching using NBLAST

  • plot neurons in 2D and 3D

  • import neurons into Blender 3D e.g. for making high quality renderings

  • export to various file formats