The FAFB dataset is documented in Zheng et al., 2018.

CATMAID is a platform for collaboratively reconsructing and annotating neurons in large EM datasets. Basic instructions for viewing and tracing are available on the CATMAID page. Tools and scripts for using CATMAID to work with FAFB are archived on this GitHub repository.

The v14 instance of FAFB refers to the 14th round of alignment. This alignment was posted to a public server at Johns Hopkins. It was also used as the basis for establishing a workspace within CATMAID for multiple labs to work together. This workspace can only be joined by application, and it requires the explicit consent of all groups already in the workspace and so it it called the ‘walled garden’. Often, however, the terms v14 and ‘walled garden’ are used synonymously, because essentially everybody working in v14 is working in the walled garden.

The walled garden login is ‘fly’, password ‘superfly’. Skeletons and annotations are visible to registered users, who are listed here. You must add your contact info to this spreadsheet if you are a registered user, so that other users can figure out who you are and what lab you’re affiliated with. Walled garden access privileges and FAFB community guidelines are managed by Marta Costa, with a steering committee consisting of Davi Bock, Greg Jefferis and Michael Reiser. Registered users use the FAFB Slack channel to swap code and manage edits that require inter-lab coordination. If you are a registered user, you are expected to check the Slack channel regularly and respond promply to queries about your neurons. If you stop using the walled garden, you should delegate the task of responding to queries to another user within our lab.

Peter Li and colleagues at Google used the v14 instance to automatically segment and skeletonize neurons, as described in Li et al. 2020. Guidlines for importing skeletons from the Google autosegmentation instance into v14 are here.

The FAFB Slack channel contains an App called ‘catbot’ that uses the NBLAST algorithm to search FAFB skeletons against FlyCircuit neurons. Check the pinned posts in #general and #code. It uses a bridging registration from FAFB to FCWB (the space of the Flycircuit neurons) and elmr. To get started, consult the Readme.