id: https://sniff.world/schema/taxon-scope name: sniff_taxon_scope title: "Sniff TaxonScope: every identifier field declares the species it is scoped to" version: 0.1.0 license: https://creativecommons.org/licenses/by/4.0/ description: >- LinkML root for the species membrane over the data planes. A knowledge graph that spans dog, human, mouse, cat and horse moves identifiers across species at every join, and an identifier that crosses without its taxon becomes a claim about the wrong animal. Three Drosophila annotation ids once sat in the gene field of canine disease rows and rendered on breed pages as the gene dogs carry a change in; no check refused them because no field had ever said which species it belonged to. This schema makes that declaration explicit. A TaxonScopeManifest names every species-scoped field in the shipped data planes, the taxon it is scoped to, the kind of identifier it holds, and optionally the catalog the value must resolve in. IdentifierNamespace rows give each (kind, taxon) pair the shape its identifiers take, so a value that matches another species' namespace is refused as a leak, which is a louder failure than an unknown value. The build reads the manifest (web/scripts/validate-taxon-scope.cjs); this file is the contract the manifest and the gate obey. Same habit as the CoverageFrame family: the schema is what the build reads, never documentation about it. prefixes: linkml: https://w3id.org/linkml/ sniff: https://sniff.world/schema/ NCBITaxon: http://purl.obolibrary.org/obo/NCBITaxon_ dcterms: http://purl.org/dc/terms/ default_prefix: sniff default_range: string imports: - linkml:types see_also: - https://sniff.world/schema/coverage-frame keywords: - species scope - identifier namespace - cross-species join - membrane classes: TaxonScopeManifest: description: >- The whole membrane: the namespaces identifiers may take, the catalogs a value may be required to resolve in, and every species-scoped field in the shipped data planes. An empty manifest is a defect, not a pass. tree_root: true slots: - namespaces - catalogs - fields IdentifierNamespace: description: >- The shape an identifier of one kind takes for one taxon. Used two ways: a value in a field of this kind and taxon must match it, and a value in a field of any other taxon that matches it is a cross-species leak. slots: - kind - taxon - pattern - ignore_case - example slot_usage: pattern: required: true Catalog: description: >- A place a value may be required to resolve in: an artifact and the path within it whose selected values form the allowed set. Declared only where the catalog is complete for its taxon and kind; a partial catalog would refuse true values, and a field with no complete catalog declares none and relies on the namespace arms alone. slots: - name - artifact - path - taxon - kind TaxonScopedField: description: >- One field in one shipped artifact whose values are identifiers scoped to a species. The taxon is either a constant or read per row from a sibling field named by taxon_from, for artifacts that carry several species. slots: - artifact - path - kind - taxon - taxon_from - split - resolves_in - note rules: - preconditions: slot_conditions: taxon_from: value_presence: ABSENT postconditions: slot_conditions: taxon: required: true description: A field with no sibling taxon column declares its taxon as a constant. slots: namespaces: description: The identifier namespaces known to the membrane. range: IdentifierNamespace multivalued: true inlined_as_list: true required: true catalogs: description: The catalogs a field may be required to resolve in. range: Catalog multivalued: true inlined_as_list: true fields: description: Every species-scoped field in the shipped data planes. range: TaxonScopedField multivalued: true inlined_as_list: true required: true kind: description: What sort of identifier the values are. range: IdentifierKindEnum required: true taxon: description: The NCBI Taxonomy CURIE of the species the values belong to. range: uriorcurie pattern: "^NCBITaxon:[0-9]+$" taxon_from: description: >- Name of a sibling field on the same row that carries the row's taxon CURIE, for artifacts holding more than one species. pattern: description: >- A regular expression every identifier of this kind and taxon matches. The token {taxid} is replaced with the numeric taxon id at check time, so one row can describe a taxon-suffixed namespace. ignore_case: description: Whether the pattern is matched case-insensitively. range: boolean example: description: One real identifier from this namespace, for the reader. name: description: A short handle a field can name in resolves_in. identifier: true artifact: description: Path of the shipped JSON artifact, relative to web/. required: true path: description: >- Where the values sit inside the artifact. Dotted segments; a bare * selects every value of an object, [] selects every element of an array, and @key selects an object's keys instead of its values. required: true split: description: >- A separator on which a single string value is split into several identifiers before checking, for fields that store a joined list. resolves_in: description: >- Catalog names whose union every value must resolve in, case-insensitively. Absent means the namespace arms are the only check, and the manifest says why in the note. range: Catalog multivalued: true note: description: Why this field is declared as it is, in one or two sentences. enums: IdentifierKindEnum: description: The kinds of species-scoped identifier the data planes carry. permissible_values: gene_symbol: description: A gene symbol as the field's species names it (HGNC-style for human, Ensembl or NCBI for dog). ensembl_gene: description: A stable Ensembl gene identifier, whose prefix encodes the species. omia_phene: description: An OMIA phene identifier, whose suffix after the hyphen is the taxon id. annotation_gene: description: A species-specific annotation id that is neither a symbol nor an Ensembl id, such as a FlyBase-style gene id.