Behavior and longitudinal integration
FiberPhotometry consumes behavior; it does not duplicate pose estimation, behavioral segmentation, or longitudinal learning models. This category makes those boundaries explicit and preserves enough provenance for the ecosystem to remain reproducible.
Choose the workflow
| Need | Workflow | Boundary |
|---|---|---|
| Import pose, states, events, and intervals from specialist tools | Behavioral ecosystem interoperability | DeepLabCut, SLEAP, Keypoint-MoSeq, and BORIS remain upstream |
| Round-trip standardized pose in NWB | Native ndx-pose interoperability | unsupported video/device links remain named omissions |
| Put independent device clocks on one time coordinate | Clock synchronization | explicit matched pulses only; no implicit interpolation |
| Check whether session summaries are comparable | Across-session comparability | incompatible sessions are reported, not pooled |
| Model learning or change across sessions | Longitudinal behavior with Unspool | Unspool owns longitudinal behavior models |
Coverage gaps this category exposes
- validated direct adapters for additional annotation and pose ecosystems;
- uncertainty propagation from pose/state estimation into photometry models;
- richer standardized NWB links among video, pose, behavior, and photometry;
- drift-aware synchronization beyond affine pulse matching; and
- stable cross-package schemas for longitudinal model outputs.
The worked behavior-tool interoperability tutorial shows the handoff. Interval transformation remains auditable through the bout-policy workflow.