Skip to content

Roadmap

Phase 0 — scientific foundation

  • [x] Canonical labelled recording prototype
  • [x] Robust and OLS reference-fit prototype with provenance
  • [x] Event alignment retaining individual events
  • [x] Initial landscape and extraction audits
  • [x] Ground-truth simulation seed
  • [x] Freeze benchmark scenarios and acceptance thresholds in-repository
  • [ ] Confirm package name with likely users before first public release

Phase 1 — trustworthy preprocessing

  • [x] Channel-level validity, dropout, repeated-extreme, and flat-step QC
  • [x] Sampling-rate and gap diagnostics
  • [x] Resampling with explicit interpolation provenance
  • [x] Session-adaptive median-rate regularization and gap diagnostics
  • [x] Frozen smooth-signal irregular-clock fidelity benchmark
  • [x] Sharp-transient, missing-run, and event-boundary resampling benchmark
  • [x] Outcome-blind real-data event-coverage and selection audit across 383 IBL sessions
  • [x] New held-out real-data regularized-AsLS comparison (retained gate failure; no promotion)
  • [x] Filter API with edge-effect reporting
  • [ ] Bleaching models and control-free baseline comparators
  • [x] Experimental double-exponential and AsLS signal-only APIs
  • [x] Frozen v0.1 benchmark with retained partial failure
  • [x] v0.2 baseline-fidelity, normalization, and sampling-rate benchmark
  • [x] First independent real-data-control pilot (mixed; no promotion)
  • [ ] Broader independent-control validation before typed-pipeline promotion
  • [x] Event-correlated reference diagnostic benchmarked against confounds
  • [ ] Reliable reference-lag diagnostic (v0.3 derivative scan failed)
  • [x] Wavelength-aware hemodynamic correction and spectral-unmixing contract
  • [x] Typed components, channel wavelengths, coefficients, offsets and calibration identity
  • [x] Independent known-component calibration with rank and conditioning gates
  • [x] Pointwise missing-channel-pattern identifiability without interpolation
  • [x] Reconstruction residuals and overdetermined channel-holdout diagnostics
  • [ ] Raw-system calibration and biological-recording validation across two sensor families
  • [x] Optogenetic-pulse artifact masks and recovery-window diagnostics
  • [x] Open, versioned sensor registry with kinetic, saturation, and interpretation constraints
  • [x] Sensor-kinetic forward models and guarded deconvolution
  • [x] Versioned model/profile/calibration identity without converting descriptive kinetics
  • [x] Difference-of-exponentials and sampled empirical causal kernels
  • [x] Gap-bounded forward prediction with explicit zero-state boundary evidence
  • [x] Prospective sampling, duration, transfer and regularization identifiability gates
  • [x] Conditional unconstrained/nonnegative recovery with complete reconstruction evidence
  • [ ] Cross-regularization robustness summary and calibrated latent uncertainty
  • [ ] Independent calibration fixtures across two sensors and realistic model mismatch
  • [x] Initial three-scenario benchmark report with retained failure
  • [x] Seven-scenario v0.2 benchmark with retained failures
  • [x] Twelve-session, four-animal IBL channel-QC audit
  • [x] Initial raw/peri-event/corrected diagnostic plot API

Phase 2 — interoperability

  • [x] Core NWB round trip and ndx-fiber-photometry response-series read path
  • [x] DANDI 001084 bounded remote-stream integration fixture
  • [x] DANDI archived dF/F numerical reproduction and provenance discrepancy report
  • [x] DANDI 001084 API metadata contract and bounded-streaming plan
  • [x] Pinned DANDI 000971 raw calcium/isobestic adapter and frozen pilot
  • [x] DANDI 000351 raw-to-processed parity audit (retained failure)
  • [ ] Recover DANDI 000351 raw-to-dF/F transformation provenance
  • [x] IBL table adapter with alternating-wavelength interpolation and masks
  • [x] Real IBL session reproduction against archived analysis outputs
  • [x] TDT and generic tabular adapters
  • [x] Schema-first wide CSV/TSV recording and event adapter
  • [x] Explicit TDT stream/epoc adapter through the same canonical boundary
  • [x] Checksum-pinned official real-block TDT integration fixture
  • [x] Native Doric, Neurophotometrics, and pyPhotometry adapters
  • [x] Shared acquisition inspection, provenance, and conformance boundary
  • [x] Doric HDF5 series/digital mapping with official-file parity
  • [x] Neurophotometrics Flags/LedState alternating-wavelength import
  • [x] pyPhotometry legacy and v1.1 pulsed binary import
  • [x] Behavioral ecosystem adapters
  • [x] Typed pose, continuous-covariate, point-event and interval boundaries
  • [x] Native-shaped DeepLabCut, SLEAP, Keypoint-MoSeq and BORIS adapters
  • [x] Executable cross-tool composition tutorial and explicit gap register
  • [x] Versioned real-file fixtures and numerical-parity matrix
    • [x] Official checksum-pinned SLEAP legacy Analysis HDF5 fixture
    • [x] Official checksum-pinned BORIS tabular CSV fixture
    • [x] Current DeepLabCut single- and multi-animal writer-contract fixtures
    • [x] Current Keypoint-MoSeq results.h5 writer-contract fixture
    • [x] Current SLEAP standard-preset and BORIS aggregated fixtures
  • [x] Direct loss-aware ndx-pose 0.3 NWB inspection, 2D/3D import, export, and round trip
  • [x] Typed affine clock-synchronization evidence, drift diagnostics and refusal thresholds
  • [x] Versioned metadata completeness and analysis/NWB/publication readiness report

Phase 3 — inference

  • [x] Typed, versioned experimental-design representation
  • [x] Experimental animal-level hierarchical bootstrap
  • [x] Experimental design-aware sign-flip and label permutations
  • [x] Frozen pseudoreplication benchmark contrasting trial and animal resampling
  • [x] Initial independent MixedLM point-estimate parity
  • [x] Independent interval and unbalanced mixed-model plumbing parity
  • [x] Extended non-Gaussian, heteroscedastic, unequal-count calibration
  • [x] Condition-stratified hierarchical resampling
  • [x] Four-animal IBL event-table/design integration
  • [x] Animal-level Welch and paired-t interval comparators
  • [x] Frozen conditional power grid from 6–30 animals per condition
  • [x] Conservative design-aware scalar inference recommender
  • [x] Independent SciPy parity for Welch and paired intervals
  • [x] Versioned analysis plans with explicit assumption acknowledgement
  • [x] Result provenance with package version, timestamp, input fingerprint, and seed
  • [x] Seeded Monte Carlo plan execution
  • [x] First frozen descriptive IBL analysis plan and result
  • [x] Typed preprocessing-to-inference pipeline with non-destructive QC gates
  • [x] Typed multiverse expansion with stable IDs and compatibility rules
  • [x] Failure-retaining robustness and decision summaries
  • [x] Reference-pipeline leave-one-animal-out diagnostics
  • [x] First frozen descriptive public-data multiverse
  • [x] Specification-curve plot API and frozen IBL figure
  • [x] Prospective, new-animal-gated IBL expansion protocol
  • [x] Refresh held-out IBL manifest (gate failed: no labelled reference channels)
  • [x] Resolve new-cohort IBL channel provenance (470-nm-only acquisition)
  • [x] Implement the published rolling baseline with 20/50-Hz and gap fixtures
  • [x] Freeze signal-only IBL v0.3 (18 animals; 15 executable universes)
  • [x] Execute and report the amended signal-only IBL v0.3.2 multiverse
  • [x] Pilot power sensitivity ranges
  • [x] Opt-in scalar mixed-model sensitivity summaries
  • [ ] fastFMM bridge and numerical-parity fixtures
  • [x] Animal-level peri-event pointwise and simultaneous interval reporting
  • [x] Common animal-level materialization and interaction boundary across peri-event, transient, state-band-power, and multi-signal workflows
  • [ ] Behavioral/event-kernel GLMs with grouped cross-validation
  • [x] Typed Gaussian FIR model for overlapping events and continuous covariates
  • [x] Leakage-safe animal/session-held-out ridge selection
  • [x] Ground-truth recovery fixture and executable simulation tutorial
  • [x] Public-data literature reproduction with a frozen design and retained weak validation
  • [x] Conditional grouped-jackknife kernel intervals and out-of-fold residual diagnostics
  • [x] Validity-mask-aware complete-case fitting with coverage evidence and gap-safe residual diagnostics
  • [ ] Formal interval-coverage calibration and simultaneous kernel bands
    • [x] Freeze and execute the first 480-study event/progress calibration
    • [x] Retain the failed normalized-progress gate without default promotion
    • [ ] Validate a revised whole-model band across every frozen scenario
  • [x] Named whole-spec design alternatives in reproducible robustness multiverses
  • [x] Basis, trial-history, duration/amplitude, and predictor-family contribution alternatives
    • [x] Typed full-FIR and raised-cosine kernel bases with physical-lag reconstruction
    • [x] Explicit current/lagged event-value modulation with session-local history
    • [x] Duration modulation and normalized-progress kernel alternatives
    • [x] Paired held-out predictor-family contribution summaries
  • [x] Validated longitudinal neural-summary handoff to Unspool
  • [x] Frozen cross-package public IBL learning-trajectory benchmark with retained negative lagged-neural forecast
  • [x] Across-session photometry comparability diagnostics before longitudinal handoff
  • [x] Interval/bout event rules and variable-duration models
  • [x] Preserve point/state distinction and physical bout duration
  • [x] Explicit onset, offset and normalized-progress projections
  • [x] Ordered merge/split/filter rules, contextual labels, overlap policies, lineage ledger, and evidence fingerprint
  • [x] Duration/amplitude and progress-basis event-kernel alternatives
  • [ ] Single-signal time/frequency/state analysis
  • [x] Gap-aware autocorrelation and power spectral density
  • [x] Spectrograms with explicit detrending, edge, and missing-data evidence
  • [x] User-supplied state/epoch tables and animal-aware comparisons
  • [x] Exact-axis animal-level inference for complete PSD and autocorrelation curves
  • [ ] Public sleep/state or long-duration worked example
  • [x] Multiscale long-duration summaries without unvalidated tonic/phasic biological labels
  • [x] Named physical-time scales with explicit step, coverage and sample gates
  • [x] Irregular-clock time weighting plus explicitly sample-weighted alternatives
  • [x] Gap/state-bounded acceptance ledger and deterministic evidence fingerprint
  • [x] Equal-session animal inference for paired and independent conditions
  • [x] Experimental spontaneous transient detection with local-baseline sensitivity analysis
  • [x] Prospective three-animal public dLight construct validation with retained detector disagreement
  • [x] Separate candidate detection from quantification on non-z-scored dF/F
  • [x] Named GuPPY, PASTa, and prominence-compatible detector families
  • [x] Baseline/control-derived frozen thresholds, compound events, and cut-waveform QC
    • [x] Baseline/control-derived frozen threshold objects
    • [x] Compound-event group and rank metadata
    • [x] Gap-aware cut waveforms and boundary QC
  • [x] Animal-aware rate and kinetic inference
  • [ ] Raw-signal and manual-annotation validation across two sensors/acquisition systems
  • [x] Multi-site/multi-color paired and conditional association workflows
  • [x] Paired channel/site metadata, alignment, control, and crosstalk diagnostics
  • [x] Lagged and event/behavior-residualized association with blocked uncertainty
  • [x] State-conditioned coherence and phase after single-signal spectral validation
  • [x] Exact-axis animal-level inference for complete lag-association and coherence curves
  • [x] Coordinate-aware mouse-level models for dense multi-fiber arrays
    • [x] Shared coordinate space/unit and complete pair/exclusion ledger
    • [x] Physical-distance bins and within-session node-label spatial null
    • [x] Equal-session mouse-level paired and independent contrasts
    • [ ] Raw-system validation with known geometry and injected shared artifacts

Phase 4 — adoption

  • [x] Publishable MkDocs site with strict link and API-reference builds
  • [x] Scientist-task methods catalog and supported/experimental/planned matrix
  • [ ] Literature-shaped worked examples for each major method family
  • [x] Public-data evidence atlas linking figures to estimands, units, and limitations
  • [x] Event-locked public IBL analysis
  • [x] Raw-NWB animal-level robustness analysis
  • [ ] Behavioral event-kernel GLM
    • [x] Ground-truth implementation tutorial
    • [x] Public-data literature reproduction with retained negative held-out R²
    • [x] Cross-tool pose/state/annotation interoperability tutorial
  • [ ] Trial-level functional mixed model
  • [ ] Long-duration tonic/phasic and spontaneous events
  • [ ] Multi-site/multi-color association

  • [ ] Spectral or hemodynamic correction with controls

  • [x] First scientist-facing event-analysis workflow

  • [x] Self-contained HTML evidence report with animal/QC/provenance views
  • [x] Declarative TOML configuration for no-code-rewrite analysis choices
  • [x] CLI from tabular project configuration to preflight, JSON, and HTML artifacts
  • [x] CLI export of raw/processed signals, events, QC, and provenance to NWB
  • [x] Unit-safe grouped report for complete multiverse robustness results
  • [x] Configuration-driven multiverse preflight, execution, and evidence bundle
  • [x] Extend project multiverses to signal-only preprocessing recipe families
  • [x] Add per-unit-lane practical-effect thresholds and machine-readable summaries
  • [x] Add declarative method-specific baseline parameters and compatibility rules
  • [x] Add multiverse-aware NWB provenance and result export
  • [x] Add a project-level result reader for JSON and NWB evidence bundles
  • [x] Add cross-bundle comparison and reproducibility-diff reporting
  • [x] Add signed publication manifests and detached verification
  • [x] Add release/DOI deposition packaging and archival metadata validation
  • [x] Add a sandbox-first Zenodo draft upload and validation adapter
  • [x] First-class candidate-to-gated-to-complete event coverage API and report panel
  • [x] Public IBL tutorial from import to fingerprinted JSON/HTML report
  • [x] Canonical raw-NWB to animal-level robustness tutorial
  • [x] Freeze the DANDI 000971 cohort, estimand, and eight-universe protocol
  • [x] Add the executable workflow and synthetic end-to-end regression fixture
  • [x] Execute the frozen public cohort and publish the retained result narrative
  • [ ] Usability review with practicing photometry scientists
  • [x] Freeze v0.1 protocol, stimulus generator, response sheet, and scoring key
  • [ ] Run five moderated sessions and publish the de-identified synthesis
  • [ ] External reproduction by two laboratories
  • [ ] Stable schema and deprecation policy
  • [x] Declare prospective v0.1 supported and experimental API surfaces
  • [x] Version the primary JSON result and package its normative JSON Schema
  • [x] Add deprecation, migration, changelog, and security policies
  • [x] Complete clean-install and canonical-artifact release audit
  • [x] First versioned public benchmark protocol and results
  • [x] Scientific decision records and method review guidance
  • [ ] External contributor governance
  • [ ] 1.0 release and archival DOI