The “Mapping the Dynamic Synaptome” project uses DERIVA to capture, curate, and share single-synapse–resolution imaging of larval zebrafish brains, revealing region-specific synapse gain and loss during memory formation. All raw and processed data are available through the Synapse repository powered by DERIVA, with persistent DOIs for every figure and dataset.
Why DERIVA underpins the Synaptome map
| Challenge | DERIVA feature |
|---|---|
| Continuous FAIR capture of multi-TB SPIM volumes, behavioral videos, and analysis tables | “Continuous FAIRness” data-lifecycle model: files land in Hatrac object store, metadata in ERMrest, DOIs minted on ingest :contentReference[oaicite:6]{index=6} |
| Evolving schema (new imaging channels, QC metrics) | Model-driven UI (Chaise) updates automatically—no code rewrites |
| Link raw ↔ analysis ↔ publication | Each figure in the PNAS paper resolves to underlying datasets via DERIVA landing pages :contentReference[oaicite:7]{index=7} |
| Reproducible analytics | Versioned CSV exports feed Python notebooks; provenance tracked back to original OME-TIFF stacks |
Impact
- 11 learner fish, 6 partial learners, 33 control animals imaged twice at single-synapse resolution.
- ≈ 1,155 synapses per fish were detected (0.5 % of total pallial synapses) and mapped in 3-D :contentReference[oaicite:8]{index=8}.
- Ventrolateral pallium showed net 30 % synapse gain while dorsomedial pallium showed loss—first in-vivo structural map of memory at synapse scale :contentReference[oaicite:9]{index=9}.
- All datasets publicly released on synapse.isrd.isi.edu with six figure-level DOIs (e.g., 10.25551/1/1-1Z06).
Typical workflow
- Imaging session – Selective-plane illumination microscope writes OME-TIFF stacks (~20 GB/fish).
- Automated ingest – DERIVA watcher uploads files, extracts metadata, creates draft records.
- Curator review – Chaise interface confirms ROI bounds and QC metrics; synapse-detection CSV attached.
- Publication – “Release” changes ACLs, mints DOIs, and exposes data through the public Synapse portal and Google Dataset Search.
Lessons for future atlases
- FAIR at acquisition time avoids post-hoc wrangling when datasets hit terabyte scale.
- Schema-agnostic UI lets neurobiology teams iterate on QC metrics without waiting for database admins.
- Region-specific statistics can be derived directly from DERIVA queries—no bespoke ETL needed.
Reference: Dempsey WP et al. “Regional Synapse Gain and Loss Accompany Memory Formation in Larval Zebrafish,” PNAS 119 (3): e2107661119 (2022). :contentReference[oaicite:10]{index=10}