MBO is an ontology for describing mouse behaviors with operational (lab) definitions and clear mappings to external standards. It focuses on observable actions and states (e.g., locomotion, rearing, sniffing) grounded in literature.

What’s here:

• Core ontology: src/mbo-core.ttl
• Measures vocabulary (SKOS): vocab/mbo-measures.ttl
• SHACL shapes: shapes/mbo-profile.shacl.ttl
• CSV templates: templates/
• CSV → TTL generator: generate_ttl.py
• Examples: examples/
• Generated output: build/mbo-generated.ttl

• Shared definitions enable comparable, interoperable datasets.
• SHACL shapes enforce consistent annotations (e.g., states have end time, events don’t).
• SKOS mappings connect to NBO/GO/CMO when appropriate.

src/                 core ontology (OWL + some SKOS categories)
vocab/               SKOS vocabularies (measures)
shapes/              SHACL constraints
templates/           CSV templates for behaviors/measures
examples/            sample instance data
build/               generated TTL files (artifacts)
generate_ttl.py      CSV → TTL generator for behaviors

1. Define or edit behaviors in templates/mbo-behaviors.csv:

   • Behavior IRI, label
   • Parent types (e.g., mbo:StateBehavior|mbo:PrimitiveBehavior)
   • Definition, category, evidence
   • External xrefs (CURIEs like NBO:0000009|GO:0071965)

2. Generate TTL

• python3 generate_ttl.py (writes build/mbo-generated.ttl)

3. Validate with SHACL (recommended)

• Install tools: pip install rdflib pyshacl
• Validate examples against shapes and ontologies:
  • pyshacl -s shapes/mbo-profile.shacl.ttl -m -f human -e src/mbo-core.ttl -e vocab/mbo-measures.ttl -e build/mbo-generated.ttl examples/mbo-newbehaviors.ttl

Expected: Conforms: True. If not, the report explains violations (e.g., missing mbo:hasEnd for a state behavior).

• skos:exactMatch: only when IRIs are fully substitutable (rare across modeling layers).
• skos:closeMatch: near-equivalent meaning with different modeling scope.
  • Example: mbo:Locomotion ↔ NBO:0000009, GO:0071965.
  • Example: mbo:ExploratoryBehavior ↔ GO:0035640.
• skos:relatedMatch: helpful pointer across layers (e.g., measurement type vs feature term).
  • Example: mbo:TimeMoving ↔ CMO:0000954.

• CSV → TTL generation runs: build/mbo-generated.ttl is produced and syntactically tidy (no trailing semicolons).
• Ontology consistency: classes and properties referenced by shapes/examples are defined in src/mbo-core.ttl.
• Measures vocab: added obo: prefix and skos:relatedMatch to CMO where appropriate.
• Note: If you run pySHACL locally, use the command above to fully validate (this environment lacked pip tooling).

• Generator enhancements:
  • Add MappingProperty column per xref (e.g., closeMatch, relatedMatch) instead of defaulting to closeMatch.
  • Add a templates/mbo-measures.csv generator to produce a measures TTL (parallel to vocab/mbo-measures.ttl).
  • Escape and datatype handling (units): consider QUDT/OM for measures and add SHACL unit constraints.
• Ontology refinements:
  • Define supported/unsupported rearing explicitly (classes exist in core, can add to CSV when needed).
  • Consider mbo:TransitionBehavior if you later model transitions or bouts.
  • Introduce controlled individuals for experimental zones/objects to standardize mbo:hasTarget.
• Validation/automation:
  • Add a GitHub Actions workflow to run: CSV→TTL, RDF parse (rdflib), SHACL validation (pySHACL), and diff on build artifacts.
  • Add a small tests/ directory with sample graphs and expected SHACL conformance.
• Documentation:
  • Add a short “data modeling examples” section showing typical annotations for each primitive behavior.
  • Add license (suggested: CC-BY 4.0 for ontology; Apache-2.0 or MIT for code).

• Propose new behaviors/measures via CSVs; include PubMed/PMC evidence.
• Use SKOS mappings conservatively (closeMatch vs relatedMatch per guidance above).
• Open PRs with clear rationale and references.