Project and use
Task, intended use, data unit and inclusion or exclusion scope.
LUMESAGE / EXPERT DATA
Biological data specifications: task to acceptance
This is a project-specification structure, not dataset inventory, real samples or a completed project report. Field applicability, values and thresholds require project agreement.
Six field groups to agree
Provenance and use rights
Source type, permission verification status, permitted uses, delivery and redistribution limits.
Experiments and samples
Species or sample type, acquisition or experimental conditions, batches and controls; explain non-applicable fields.
Labels and evidence
Label definitions, evidence type, protocol version, expert review, disagreements and uncertainty.
Splits and evaluation
Grouping unit, deduplication, independent validation design, scoring version and limitations.
Acceptance and versions
Quality metrics, agreed thresholds, sampling coverage, rework conditions, file manifest and change history.
Make better data an acceptance specification
Scientific context
Record provenance, sampling unit, experimental conditions, batches and controls. Identify observations, expert judgments and model predictions.
Engineering quality
Check field completeness, units, formats, versions and permissions. Maintain issue and revision logs.
Model validation
Choose grouping, deduplication and leakage checks for the task. Record baselines, scoring versions, failure types and limitations.
Agree quality thresholds, sampling coverage, rework rules and timelines before the pilot. A generic accuracy claim does not replace task-specific acceptance.
Download the machine-readable field specification
The JSON Schema validates field structure. It does not establish permissions, scientific validity or completion of expert review.
JSON Schema ↓