Decentralized platform for connectomics data
Store, query, and share synapse metrics from electron microscopy datasets. Cloud-native, federated, and open to the community.
Store, query, and share synapse metrics from electron microscopy datasets. Cloud-native, federated, and open to the community.
A cloud-native data platform for connectomic research that can be openly queried and externally contributed to, facilitating data access and collaboration across experimental and computational neuroscience.
Use SyQL, a structured query language designed for connectomics data, to search geometric, categorical, and relational data types across neuronal compartments.
Deploy on Kubernetes clusters and form a data federation across voluntary data centers. Contributors retain full data ownership with cite-able DOIs.
Perform graph analysis, meta-analysis, and analytics on synapse, axon, dendrite, vesicle, and mitochondria datasets at scale.
Built on a modern stack for performance, scalability, and accessibility.
PostgreSQL, ClickHouse, MinIO, and Apache Arrow Flight for fast columnar queries and object storage.
Contributors retain ownership, associate cite-able DOIs per dataset, and can modify or update their data at any time.
Integrate any number of voluntary data centers connected through the SynDB API endpoint for distributed queries.
Web UI, REST API, and CLI. Access your data however you prefer, from browser to terminal.
Automated ingestion from major connectomics data sources with validation, transformation, and graph precomputation.
Kubernetes-native deployment with ClickHouse sharding for scaling across data centers and workloads.
SynDB is developed with infrastructure partnership, institutional support, and research funding from organizations advancing open neuroscience.
Curated collection of synaptic connectome and neuro-anatomical datasets across model organisms and human tissue.
SynDB’s ClickHouse backend has moved to the post-P09 distributed_staged architecture: a 5-shard, 2-replica cluster that now serves as the sole authoritative deployment for production data. The legacy single-node staging deployment was retired during the cutover window, with the irreversible standalone cleanup completed on 2026-06-01.
SyQL queries now fan out across the distributed cluster instead of a single node, giving larger datasets a clear path for shard-local query execution and future growth.
Base tables now run with replica-backed storage inside each shard, removing the old single-node failure domain while preserving the existing nightly backup cadence.
The 5 × 2 layout adds room for the next dataset class without a topology reset, with per-shard storage and compute scaling independently as the corpus grows.
The public API and UI remained online through the transition, so users kept the same entry points while the storage layer moved underneath.
Explore the platform, read the documentation, or dive into the source code.
