CAVE: Connectome Annotation Versioning Engine.

Area of Science:

• Neuroscience
• Computational Biology
• Bioinformatics

Background:

• Large-scale connectomic datasets are growing due to advances in electron microscopy and computational infrastructure.
• Sharing and collaborative annotation of these datasets are crucial for neuroscience research.
• Concurrent annotation and error correction (proofreading) are necessary for accurate cell identity labeling in massive datasets.

Purpose of the Study:

• To present the Connectome Annotation Versioning Engine (CAVE) as a computational infrastructure.
• To enable immediate and reproducible connectome analysis on petascale datasets.
• To support ongoing proofreading and annotation processes.

Main Methods:

• CAVE provides a distributed infrastructure for continuous versioning of large neuronal reconstructions.
• Annotations are location-defined, enabling rapid assignment to underlying segments for efficient analysis.
• The system supports schematized and extensible annotations for diverse research needs.

Main Results:

• CAVE facilitates immediate and reproducible analysis of petascale connectomic datasets.
• It allows concurrent proofreading and annotation, managing millions of voxel relabelings and thousands of annotations.
• Fast analysis queries are possible across arbitrary time points of the evolving dataset.

Conclusions:

• CAVE offers a robust solution for managing and analyzing large, dynamic connectomic data.
• It supports collaborative research by providing versioned access to constantly updated neuronal reconstructions.
• The infrastructure is scalable and already implemented for some of the largest connectomics datasets available.