@prefix sh: . @prefix genomics: . @prefix cascade: . @prefix prov: . @prefix xsd: . @prefix rdfs: . # ============================================================================ # Cascade Protocol — Genomics Vocabulary SHACL Validation Shapes # ============================================================================ # Version: 1.0-draft (TASK-0.2, 2026-05-05; +v1-draft.0.2 additions, 2026-05-05) # Validates against: genomics.ttl v1.0-draft (genomics@v1-draft.0.2) # # Changelog: # v1-draft.0.2 (2026-05-05): added VariantShape property shapes for the # four new Variant-domain properties — genomics:refAllele, altAllele, # genomicStartEnd (Info-severity datatype/maxCount checks), # genomics:variantAlleleFrequency (Violation-severity 0.0–1.0 range), # genomics:somaticStatus (Violation-severity sh:in over Germline / # Somatic / UnknownSomaticStatus). All optional (no minCount). No # NodeShape added for the new generic genomics:reportedRecord predicate # — it is intentionally domain-broad on GeneticTest with no rdfs:range, # so a SHACL constraint would fight the design intent. Existing fixtures # never reference any of these new properties, so existing data passes # unchanged. # v1-draft.0.1 (2026-05-05): initial 12 NodeShapes (TASK-0.2). # # Shapes authored (12 sh:NodeShape): # 1. genomics:VariantShape — required identity, stable IDs, # quality tier # 2. genomics:VariantInterpretationShape — D-Q5 cardinality, ACMG enum, # D-QUALITY-TIER safety constraint # 3. genomics:HaplotypeShape — required star-allele identity # 4. genomics:DiplotypeShape — exactly two haplotypes (hapA+hapB) # 5. genomics:CopyNumberVariantShape — interval + copy number, VAF range # 6. genomics:SubmitterAssertionShape — submitter + asserted class # 7. genomics:PedigreeShape — proband + at least one member # 8. genomics:PedigreeMemberShape — required relationship role # 9. genomics:GeneticTestShape — required test type # 10. genomics:GeneticTestOrderShape — required order status + date # 11. genomics:SequencingRunShape — coverage, technology, reference # 12. genomics:RawFileShape — format, hash, location, reference # # SEVERITY LEVELS (matching advisory.shapes.ttl / health.shapes.ttl) # sh:Violation — Must fix; data is invalid or unsafe without it. # (Required-field absences and the D-QUALITY-TIER safety # constraint use this severity.) # sh:Warning — Should address; data is technically usable but missing # recommended completeness or audit fields. # sh:Info — Suggested enrichment; nice-to-have. # # ---------------------------------------------------------------------------- # D-Q5 — Multi-condition VariantInterpretation cardinality # ---------------------------------------------------------------------------- # Each VariantInterpretation is per-(variant, condition) pair: SHACL enforces # genomics:variantInterpreted minCount/maxCount = 1 and genomics:condition # minCount/maxCount = 1 on VariantInterpretationShape. Multi-condition cases # produce multiple Interpretation instances (one per condition); same-pair # reclassifications produce new Interpretation instances chained via # prov:wasRevisionOf. # # The reclassification-chain integrity rule (a revision MUST share both # variantInterpreted and condition with its predecessor) is NOT expressible # in pure SHACL Core without sh:sparql, because the constraint is across two # focus nodes (the revising and revised interpretations). This is documented # here as a cross-class invariant; the supplementing code-level check lives # in cascade-cli (TASK-4.x — applier validation step on incoming # VariantReclassification advisories). Adding a sh:sparql block here was # considered and rejected as making the shape harder to read for the # narrow benefit; the cascade-cli check is the authoritative enforcer. # # ---------------------------------------------------------------------------- # D-QUALITY-TIER — Safety constraint on Pathogenic / LikelyPathogenic # ---------------------------------------------------------------------------- # This is THE critical constraint of this shape file. For every # VariantInterpretation whose acmgClassification is Pathogenic or # LikelyPathogenic, EITHER: # (a) the referenced genomics:Variant has # genomics:dataQualityTier genomics:ClinicalGrade, OR # (b) the Interpretation itself has genomics:requiresConfirmation true. # # Rationale: prevents consumer-grade pathogenic calls from masquerading as # clinical findings. Phase 2C (consumer-array imports — 23andMe, AncestryDNA, # MyHeritage, FTDNA) auto-tags every variant ConsumerGrade and would, without # this constraint, be able to push Pathogenic interpretations into the # advisory engine. The advisory engine MUST NOT auto-apply care # recommendations to confirmation-required interpretations regardless of # issuer trust. # # Expression: VariantInterpretationShape uses sh:xone (exactly-one of two # disjoint conditions): the qualifying-class predicate sh:not [sh:in # (Pathogenic LikelyPathogenic)] (i.e., classification is benign-leaning or # VUS — constraint trivially satisfied), OR a require-mitigation predicate # combining (variant.dataQualityTier = ClinicalGrade) with # (requiresConfirmation = true) using sh:or. Using sh:or at the inner level # is correct because either mitigation is sufficient. Using sh:xone at the # outer level is the readable way to encode "if-then" in pure SHACL Core # (one of: classification is non-pathogenic, or one of the mitigations # applies). The violation message names the affected interpretation IRI # and lists both remediation paths so callers know what to change. # # ---------------------------------------------------------------------------- # Vocabulary gaps acknowledged in shape comments (not enforced) # ---------------------------------------------------------------------------- # - genomics:GeneticTest has no testDate property in v1-draft.0.1; only # genomics:GeneticTestOrder has genomics:orderedAt. Test-result date is # captured indirectly via prov:generatedAtTime on the GeneticTest # instance. GeneticTestShape requires testType but not a test date; # date enforcement deferred to v1-draft.0.2 if a property is added. # - genomics:SubmitterAssertion has no assertionDate property and no # forward reference to its parent VariantInterpretation; the inverse # link is genomics:aggregatedFrom (from Interpretation to Assertion). # SubmitterAssertionShape therefore requires submitter and asserted # classification but cannot directly enforce a parent-interpretation # reference. This is acceptable because Interpretation→Assertion is the # authoritative direction in the model. # ============================================================================ # ============================================================================ # Shape 1: Variant # ============================================================================ genomics:VariantShape a sh:NodeShape ; sh:targetClass genomics:Variant ; rdfs:label "Variant Shape"@en ; rdfs:comment "Validation constraints for genomics:Variant. Enforces required gene-symbol identity, presence of at least one stable identifier across the four supported registries (CAid, VRS, ClinVar, dbSNP), and presence of a data-quality tier (used by the D-QUALITY-TIER safety constraint on downstream VariantInterpretations)."@en ; # WARNING: geneSymbol recommended (HGNC approved). Most clinical-grade # imports carry it, but a Variant remains valid via other identifiers # (CAid, VRS, ClinVar VCV, dbSNP rs, or the refAllele+altAllele+ # genomicStartEnd coordinate triple). VRS preserve-only imports # (D-Q6) and gene-less VCF records legitimately lack gene context; # downgraded from Violation to Warning in v1-draft.0.3 per the # 2026-05-06 VRS test-fixture tie-break (cascade-coordination/ # tie-breaks/2026-05-06-vrs-geneSymbol-shape.md). When present, the # symbol must be exactly one non-empty string. sh:property [ sh:path genomics:geneSymbol ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:minLength 1 ; sh:severity sh:Warning ; sh:name "Gene Symbol"@en ; sh:message "Recommended: genomics:geneSymbol (HGNC approved symbol, e.g., 'BRCA2'). Variant remains valid via stable identifiers (CAid/VRS/ClinVar/dbSNP) or VCF coordinates (refAllele+altAllele+genomicStartEnd) when source data lacks gene context."@en ] ; # VIOLATION: dataQualityTier required and must be one of the four # published DataQualityTier named individuals. Required so the # D-QUALITY-TIER safety constraint on VariantInterpretation can fire # against a non-empty tier value. sh:property [ sh:path genomics:dataQualityTier ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( genomics:ClinicalGrade genomics:ResearchGrade genomics:ConsumerGrade genomics:UnknownQuality ) ; sh:severity sh:Violation ; sh:name "Data Quality Tier"@en ; sh:message "Variant must declare genomics:dataQualityTier as one of the four published genomics:DataQualityTier named individuals (ClinicalGrade, ResearchGrade, ConsumerGrade, UnknownQuality). Required so the D-QUALITY-TIER safety constraint on downstream VariantInterpretations can be evaluated."@en ] ; # Stable-identifier datatype constraints (per-property). The cross-class # 'at least one stable ID' constraint is enforced by sh:or at the # NodeShape level below. sh:property [ sh:path genomics:caId ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "ClinGen Allele Registry CAid"@en ] ; sh:property [ sh:path genomics:vrsId ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "GA4GH VRS Identifier"@en ] ; sh:property [ sh:path genomics:clinvarVariationId ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "ClinVar Variation ID"@en ] ; sh:property [ sh:path genomics:dbsnpRsId ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "dbSNP rsID"@en ] ; # VIOLATION: at least one stable identifier among caId / vrsId / # clinvarVariationId / dbsnpRsId. HGVS strings are descriptive but # normalization-fragile — a stable ID is what joins reliably across # databases. sh:or evaluates each branch; at least one must report # zero violations. sh:or ( [ sh:path genomics:caId ; sh:minCount 1 ] [ sh:path genomics:vrsId ; sh:minCount 1 ] [ sh:path genomics:clinvarVariationId ; sh:minCount 1 ] [ sh:path genomics:dbsnpRsId ; sh:minCount 1 ] ) ; # WARNING: zygosity recommended for variant-level reasoning. sh:property [ sh:path genomics:zygosity ; sh:maxCount 1 ; sh:severity sh:Warning ; sh:name "Zygosity"@en ; sh:message "genomics:zygosity is recommended; absence forces downstream tools to assume Heterozygous"@en ] ; # WARNING: mosaicismFraction range check (0.0..1.0 inclusive) when present. sh:property [ sh:path genomics:mosaicismFraction ; sh:datatype xsd:decimal ; sh:maxCount 1 ; sh:minInclusive 0.0 ; sh:maxInclusive 1.0 ; sh:severity sh:Violation ; sh:name "Mosaicism Fraction (VAF)"@en ; sh:message "genomics:mosaicismFraction, if present, must be an xsd:decimal between 0.0 and 1.0 inclusive (variant allele fraction)"@en ] ; # Added in v1-draft.0.2: VAF as a distinct property from mosaicismFraction. # Range 0.0–1.0 inclusive. Optional — no minCount. sh:property [ sh:path genomics:variantAlleleFrequency ; sh:datatype xsd:decimal ; sh:maxCount 1 ; sh:minInclusive 0.0 ; sh:maxInclusive 1.0 ; sh:severity sh:Violation ; sh:name "Variant Allele Frequency"@en ; sh:message "genomics:variantAlleleFrequency, if present, must be an xsd:decimal between 0.0 and 1.0 inclusive (fraction of reads supporting the alternate allele)"@en ] ; # Added in v1-draft.0.2: VCF-style coordinate properties. All optional. sh:property [ sh:path genomics:refAllele ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "Reference Allele"@en ] ; sh:property [ sh:path genomics:altAllele ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "Alternate Allele"@en ] ; sh:property [ sh:path genomics:genomicStartEnd ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "Genomic Start-End Coordinates"@en ] ; # Added in v1-draft.0.2: somatic status (germline / somatic / unknown). # Optional but value must be one of three published named individuals. sh:property [ sh:path genomics:somaticStatus ; sh:maxCount 1 ; sh:in ( genomics:Germline genomics:Somatic genomics:UnknownSomaticStatus ) ; sh:severity sh:Violation ; sh:name "Somatic Status"@en ; sh:message "genomics:somaticStatus, if present, must be one of the three published genomics:SomaticStatus named individuals (Germline, Somatic, UnknownSomaticStatus)"@en ] . # Cross-class node-level violation message for the 'at least one stable ID' # constraint. SHACL Core sh:or does not directly produce a top-level message # tied to the parent shape, so we attach a message via sh:message on the # parent NodeShape — most engines surface it when the sh:or branch fails. genomics:VariantShape sh:message "Variant must carry at least one stable identifier among genomics:caId, genomics:vrsId, genomics:clinvarVariationId, or genomics:dbsnpRsId. HGVS strings alone are insufficient for reliable cross-database joins."@en . # ============================================================================ # Shape 2: Variant Interpretation (D-Q5 + D-QUALITY-TIER) # ============================================================================ genomics:VariantInterpretationShape a sh:NodeShape ; sh:targetClass genomics:VariantInterpretation ; rdfs:label "Variant Interpretation Shape"@en ; rdfs:comment "Validation constraints for genomics:VariantInterpretation. Implements D-Q5 (each Interpretation is per-(variant, condition); cardinality 1..1 on both genomics:variantInterpreted and genomics:condition) and D-QUALITY-TIER (the safety constraint on Pathogenic / LikelyPathogenic classifications)."@en ; # ──────────────────────────────────────────────────────────────────── # D-Q5: variantInterpreted cardinality 1..1. # Range widened in v1-draft.0.3 from genomics:Variant alone to # {Variant, CopyNumberVariant, Haplotype} via sh:or — clinical # interpretations attach to all three molecular-record types # (e.g., the retinoblastoma phenopacket interprets a chr13 CNV). # ──────────────────────────────────────────────────────────────────── sh:property [ sh:path genomics:variantInterpreted ; sh:or ( [ sh:class genomics:Variant ] [ sh:class genomics:CopyNumberVariant ] [ sh:class genomics:Haplotype ] ) ; sh:minCount 1 ; sh:maxCount 1 ; sh:severity sh:Violation ; sh:name "Variant Interpreted"@en ; sh:message "VariantInterpretation must reference exactly one Variant / CopyNumberVariant / Haplotype via genomics:variantInterpreted (D-Q5 — cardinality 1..1; multi-variant cases produce multiple Interpretation instances)"@en ] ; # ──────────────────────────────────────────────────────────────────── # D-Q5: condition cardinality 1..1 # ──────────────────────────────────────────────────────────────────── sh:property [ sh:path genomics:condition ; sh:minCount 1 ; sh:maxCount 1 ; sh:severity sh:Violation ; sh:name "Associated Condition"@en ; sh:message "VariantInterpretation must reference exactly one condition via genomics:condition (D-Q5 — cardinality 1..1; multi-condition cases produce multiple Interpretation instances, one per condition)"@en ] ; # ──────────────────────────────────────────────────────────────────── # ACMG classification — required, closed enumeration over the 5 # published AcmgClass named individuals. # ──────────────────────────────────────────────────────────────────── sh:property [ sh:path genomics:acmgClassification ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( genomics:Pathogenic genomics:LikelyPathogenic genomics:VUS genomics:LikelyBenign genomics:Benign ) ; sh:severity sh:Violation ; sh:name "ACMG Classification"@en ; sh:message "VariantInterpretation must declare genomics:acmgClassification as one of the five published genomics:AcmgClass named individuals (Pathogenic, LikelyPathogenic, VUS, LikelyBenign, Benign)"@en ] ; # ──────────────────────────────────────────────────────────────────── # Review status — optional but, when present, must be one of the seven # published ReviewStatus named individuals. # ──────────────────────────────────────────────────────────────────── sh:property [ sh:path genomics:reviewStatus ; sh:maxCount 1 ; sh:in ( genomics:NoAssertionProvided genomics:CriteriaNotProvided genomics:SingleSubmitter genomics:ConflictingSubmissions genomics:MultipleSubmittersNoConflict genomics:ExpertPanelReviewed genomics:PracticeGuideline ) ; sh:severity sh:Warning ; sh:name "Review Status"@en ; sh:message "genomics:reviewStatus, if present, must be one of the seven published genomics:ReviewStatus named individuals (NoAssertionProvided, CriteriaNotProvided, SingleSubmitter, ConflictingSubmissions, MultipleSubmittersNoConflict, ExpertPanelReviewed, PracticeGuideline). Recommended for advisory-engine trust weighting."@en ] ; # WARNING: requiresConfirmation must be xsd:boolean if present. sh:property [ sh:path genomics:requiresConfirmation ; sh:datatype xsd:boolean ; sh:maxCount 1 ; sh:severity sh:Violation ; sh:name "Requires Confirmation"@en ; sh:message "genomics:requiresConfirmation, if present, must be a single xsd:boolean"@en ] ; # WARNING: interpretedDate recommended for chronology and reclassification chains. sh:property [ sh:path genomics:interpretedDate ; sh:datatype xsd:date ; sh:maxCount 1 ; sh:severity sh:Warning ; sh:name "Interpretation Date"@en ; sh:message "genomics:interpretedDate (xsd:date) is recommended; needed for reclassification chronology"@en ] ; # ──────────────────────────────────────────────────────────────────── # D-QUALITY-TIER safety constraint # ──────────────────────────────────────────────────────────────────── # If acmgClassification is Pathogenic or LikelyPathogenic, then EITHER # (a) the referenced Variant has dataQualityTier ClinicalGrade, OR # (b) requiresConfirmation is true on the Interpretation itself. # # Encoded as sh:xone (exactly one) of two disjoint conditions: # Branch X: classification is NOT in {Pathogenic, LikelyPathogenic} # (constraint trivially satisfied, no mitigation needed) # Branch Y: classification IS in {Pathogenic, LikelyPathogenic} AND # at least one mitigation applies (ClinicalGrade variant OR # requiresConfirmation true). # # sh:xone is used so that a malformed instance with an out-of-range # acmgClassification is also caught — the acmgClassification property # constraint above (sh:in) handles the enum check; the sh:xone here # handles the conditional requirement. sh:xone ( # Branch X: classification is non-pathogenic (or absent — trapped # by the acmgClassification property constraint above). [ sh:path genomics:acmgClassification ; sh:not [ sh:in ( genomics:Pathogenic genomics:LikelyPathogenic ) ] ] # Branch Y: classification is pathogenic-leaning AND mitigation present. [ sh:and ( # Y.1: classification IS in {Pathogenic, LikelyPathogenic} [ sh:path genomics:acmgClassification ; sh:in ( genomics:Pathogenic genomics:LikelyPathogenic ) ] # Y.2: at least one mitigation applies — either the linked # variant is ClinicalGrade tier, OR the interpretation has # requiresConfirmation true. [ sh:or ( [ sh:path ( genomics:variantInterpreted genomics:dataQualityTier ) ; sh:hasValue genomics:ClinicalGrade ] [ sh:path genomics:requiresConfirmation ; sh:hasValue true ] ) ] ) ] ) ; sh:severity sh:Violation ; sh:message "D-QUALITY-TIER safety constraint violated: this VariantInterpretation declares acmgClassification Pathogenic or LikelyPathogenic, but the referenced genomics:Variant is NOT genomics:ClinicalGrade tier AND genomics:requiresConfirmation is not true on this Interpretation. To resolve: EITHER promote the variant's genomics:dataQualityTier to genomics:ClinicalGrade (only valid if the underlying sequencing is CLIA/CAP/ISO15189-certified, >=30x coverage or validated panel-equivalent, with raw files retained), OR set genomics:requiresConfirmation true on this Interpretation. Consumer-grade pathogenic calls MUST NOT auto-apply care recommendations regardless of issuer trust."@en . # ============================================================================ # Shape 3: Haplotype # ============================================================================ genomics:HaplotypeShape a sh:NodeShape ; sh:targetClass genomics:Haplotype ; rdfs:label "Haplotype Shape"@en ; rdfs:comment "Validation constraints for genomics:Haplotype. The clinically actionable identifier in pharmacogenomics (PharmVar star alleles) and HLA typing is the star-allele symbol; this is required. Component variants are recommended for full provenance but optional in v1-draft.0.1."@en ; # VIOLATION: starAlleleSymbol required — the clinically actionable PGx/HLA # identifier (e.g., 'CYP2C19*2', 'HLA-DQB1*02:01'). sh:property [ sh:path genomics:starAlleleSymbol ; sh:datatype xsd:string ; sh:minCount 1 ; sh:maxCount 1 ; sh:minLength 1 ; sh:severity sh:Violation ; sh:name "Star Allele Symbol"@en ; sh:message "Haplotype must carry exactly one non-empty genomics:starAlleleSymbol (PharmVar/HLA nomenclature, e.g. 'CYP2C19*2')"@en ] ; # WARNING: hasComponent recommended for full provenance. sh:property [ sh:path genomics:hasComponent ; sh:class genomics:Variant ; sh:severity sh:Warning ; sh:name "Component Variants"@en ; sh:message "genomics:hasComponent is recommended; absence prevents downstream tools from auditing which constituent variants define this star allele"@en ] . # ============================================================================ # Shape 4: Diplotype # ============================================================================ genomics:DiplotypeShape a sh:NodeShape ; sh:targetClass genomics:Diplotype ; rdfs:label "Diplotype Shape"@en ; rdfs:comment "Validation constraints for genomics:Diplotype. A diplotype is, by definition, a pair of haplotypes — the ontology models this with two distinct properties (genomics:hapA, genomics:hapB) rather than a single 2..2 cardinality property. Each must be present exactly once."@en ; # VIOLATION: hapA required, exactly one Haplotype. sh:property [ sh:path genomics:hapA ; sh:class genomics:Haplotype ; sh:minCount 1 ; sh:maxCount 1 ; sh:severity sh:Violation ; sh:name "Haplotype A"@en ; sh:message "Diplotype must carry exactly one genomics:hapA referencing a genomics:Haplotype"@en ] ; # VIOLATION: hapB required, exactly one Haplotype. sh:property [ sh:path genomics:hapB ; sh:class genomics:Haplotype ; sh:minCount 1 ; sh:maxCount 1 ; sh:severity sh:Violation ; sh:name "Haplotype B"@en ; sh:message "Diplotype must carry exactly one genomics:hapB referencing a genomics:Haplotype (a diplotype is by definition a pair of haplotypes)"@en ] ; # WARNING: diplotypeNotation recommended for direct match against PGx # guideline tables (CPIC, DPWG). sh:property [ sh:path genomics:diplotypeNotation ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Warning ; sh:name "Diplotype Notation"@en ; sh:message "genomics:diplotypeNotation (e.g., '*1/*2') is recommended for direct matching against CPIC/DPWG guideline tables"@en ] . # ============================================================================ # Shape 5: Copy Number Variant # ============================================================================ # Note: CopyNumberVariant rdfs:subClassOf genomics:Variant in the ontology, # so VariantShape constraints apply transitively (geneSymbol, dataQualityTier, # at-least-one-stable-ID). This shape adds CNV-specific requirements. genomics:CopyNumberVariantShape a sh:NodeShape ; sh:targetClass genomics:CopyNumberVariant ; rdfs:label "Copy Number Variant Shape"@en ; rdfs:comment "Validation constraints specific to genomics:CopyNumberVariant. Inherits VariantShape constraints (geneSymbol, dataQualityTier, at least one stable ID) via rdfs:subClassOf and adds the CNV-specific requirements: integer copy number, interval coordinates with reference."@en ; # VIOLATION: copyNumber required, integer. sh:property [ sh:path genomics:copyNumber ; sh:datatype xsd:integer ; sh:minCount 1 ; sh:maxCount 1 ; sh:minInclusive 0 ; sh:severity sh:Violation ; sh:name "Copy Number"@en ; sh:message "CopyNumberVariant must carry exactly one genomics:copyNumber as a non-negative xsd:integer (0 = homozygous deletion, 1 = heterozygous deletion, 3+ = duplication/amplification)"@en ] ; # VIOLATION: cnvIntervalRef required — interval coordinates are # uninterpretable without their reference contig. sh:property [ sh:path genomics:cnvIntervalRef ; sh:datatype xsd:string ; sh:minCount 1 ; sh:maxCount 1 ; sh:minLength 1 ; sh:severity sh:Violation ; sh:name "CNV Interval Reference"@en ; sh:message "CopyNumberVariant must carry exactly one genomics:cnvIntervalRef (reference contig name or accession, e.g., 'NC_000013.11')"@en ] ; # VIOLATION: cnvIntervalStart required, integer ≥ 1 (1-based inclusive). sh:property [ sh:path genomics:cnvIntervalStart ; sh:datatype xsd:integer ; sh:minCount 1 ; sh:maxCount 1 ; sh:minInclusive 1 ; sh:severity sh:Violation ; sh:name "CNV Interval Start"@en ; sh:message "CopyNumberVariant must carry exactly one genomics:cnvIntervalStart as a positive xsd:integer (1-based, inclusive)"@en ] ; # VIOLATION: cnvIntervalEnd required, integer ≥ 1. sh:property [ sh:path genomics:cnvIntervalEnd ; sh:datatype xsd:integer ; sh:minCount 1 ; sh:maxCount 1 ; sh:minInclusive 1 ; sh:severity sh:Violation ; sh:name "CNV Interval End"@en ; sh:message "CopyNumberVariant must carry exactly one genomics:cnvIntervalEnd as a positive xsd:integer (1-based, inclusive)"@en ] . # ============================================================================ # Shape 6: Submitter Assertion # ============================================================================ genomics:SubmitterAssertionShape a sh:NodeShape ; sh:targetClass genomics:SubmitterAssertion ; rdfs:label "Submitter Assertion Shape"@en ; rdfs:comment "Validation constraints for genomics:SubmitterAssertion (e.g., one ClinVar SCV record). Requires submitter identity and the asserted ACMG classification — these are what makes the assertion auditable. Note: the v1-draft.0.1 vocabulary does NOT define an assertionDate property nor a forward reference to the parent VariantInterpretation; the inverse link is genomics:aggregatedFrom (Interpretation→Assertion), which is the authoritative direction."@en ; # VIOLATION: submitter required. sh:property [ sh:path genomics:submitter ; sh:datatype xsd:string ; sh:minCount 1 ; sh:maxCount 1 ; sh:minLength 1 ; sh:severity sh:Violation ; sh:name "Submitter"@en ; sh:message "SubmitterAssertion must carry exactly one non-empty genomics:submitter (lab/consortium/expert-panel name, e.g., 'ENIGMA', 'Ambry Genetics')"@en ] ; # VIOLATION: assertedClassification required and must be one of the five # published AcmgClass named individuals. sh:property [ sh:path genomics:assertedClassification ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( genomics:Pathogenic genomics:LikelyPathogenic genomics:VUS genomics:LikelyBenign genomics:Benign ) ; sh:severity sh:Violation ; sh:name "Asserted ACMG Classification"@en ; sh:message "SubmitterAssertion must declare genomics:assertedClassification as one of the five published genomics:AcmgClass named individuals"@en ] ; # WARNING: scvAccession recommended for ClinVar-sourced records. sh:property [ sh:path genomics:scvAccession ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Warning ; sh:name "SCV Accession"@en ; sh:message "genomics:scvAccession is recommended for ClinVar-sourced submissions (audit trail to the original SCV record)"@en ] ; # INFO: submitterCategory recommended for review-status weighting. sh:property [ sh:path genomics:submitterCategory ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "Submitter Category"@en ; sh:message "genomics:submitterCategory is suggested; drives review-status weighting in the aggregate"@en ] . # ============================================================================ # Shape 7: Pedigree # ============================================================================ genomics:PedigreeShape a sh:NodeShape ; sh:targetClass genomics:Pedigree ; rdfs:label "Pedigree Shape"@en ; rdfs:comment "Validation constraints for genomics:Pedigree. A pedigree must have exactly one proband (the index patient who brought the family to attention) and at least one member (the proband counts)."@en ; # VIOLATION: proband required, exactly one PedigreeMember. sh:property [ sh:path genomics:proband ; sh:class genomics:PedigreeMember ; sh:minCount 1 ; sh:maxCount 1 ; sh:severity sh:Violation ; sh:name "Proband"@en ; sh:message "Pedigree must reference exactly one genomics:proband (the index patient who brought the family to clinical attention)"@en ] ; # VIOLATION: hasMember required, at least one PedigreeMember (the proband # counts as a member). sh:property [ sh:path genomics:hasMember ; sh:class genomics:PedigreeMember ; sh:minCount 1 ; sh:severity sh:Violation ; sh:name "Pedigree Members"@en ; sh:message "Pedigree must reference at least one genomics:hasMember (the proband should be included)"@en ] . # ============================================================================ # Shape 8: Pedigree Member # ============================================================================ genomics:PedigreeMemberShape a sh:NodeShape ; sh:targetClass genomics:PedigreeMember ; rdfs:label "Pedigree Member Shape"@en ; rdfs:comment "Validation constraints for genomics:PedigreeMember. The relationship-to-proband role is required and drawn from the closed HL7 v3 RoleCode-aligned named individuals declared in genomics.ttl."@en ; # VIOLATION: relativeRole required and must be one of the published # RelationshipRole named individuals (HL7 v3 RoleCode-aligned). sh:property [ sh:path genomics:relativeRole ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( genomics:Proband genomics:MTH genomics:FTH genomics:SIS genomics:BRO genomics:DAU genomics:SON genomics:MAUNT genomics:MUNCLE genomics:PAUNT genomics:PUNCLE genomics:MGRMTH genomics:MGRFTH genomics:PGRMTH genomics:PGRFTH genomics:MCOUSN genomics:PCOUSN genomics:NIECE genomics:NEPHEW ) ; sh:severity sh:Violation ; sh:name "Relationship Role"@en ; sh:message "PedigreeMember must declare genomics:relativeRole as one of the published HL7 v3 RoleCode-aligned RelationshipRole named individuals (Proband, MTH, FTH, SIS, BRO, DAU, SON, MAUNT, MUNCLE, PAUNT, PUNCLE, MGRMTH, MGRFTH, PGRMTH, PGRFTH, MCOUSN, PCOUSN, NIECE, NEPHEW)"@en ] ; # WARNING: relativeSex recommended for inheritance reasoning. sh:property [ sh:path genomics:relativeSex ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:in ("male" "female" "unknown") ; sh:severity sh:Warning ; sh:name "Biological Sex"@en ; sh:message "genomics:relativeSex is recommended (one of 'male', 'female', 'unknown'); needed for X-linked / Y-linked inheritance reasoning"@en ] ; # WARNING: carrierStatus, when present, must be one of the published values. sh:property [ sh:path genomics:carrierStatus ; sh:maxCount 1 ; sh:in ( genomics:PositiveAffected genomics:PositiveUnaffected genomics:Negative genomics:NotTested genomics:Obligate ) ; sh:severity sh:Violation ; sh:name "Carrier Status"@en ; sh:message "genomics:carrierStatus, if present, must be one of the five published genomics:CarrierStatus named individuals (PositiveAffected, PositiveUnaffected, Negative, NotTested, Obligate)"@en ] . # ============================================================================ # Shape 9: Genetic Test # ============================================================================ # Note: v1-draft.0.1 does NOT declare a testDate property on GeneticTest. # Test-result date is captured indirectly via prov:generatedAtTime on the # instance. Date enforcement deferred to v1-draft.0.2 if/when a property is # added to the ontology. genomics:GeneticTestShape a sh:NodeShape ; sh:targetClass genomics:GeneticTest ; rdfs:label "Genetic Test Shape"@en ; rdfs:comment "Validation constraints for genomics:GeneticTest (FHIR DiagnosticReport-aligned). Requires test type. Note: testDate is not yet a declared property in v1-draft.0.1; chronology comes from prov:generatedAtTime."@en ; # VIOLATION: testType required and must be one of the published TestType # named individuals. sh:property [ sh:path genomics:testType ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( genomics:SingleGeneTest genomics:GenePanelTest genomics:ExomeSequencing genomics:GenomeSequencing genomics:ChromosomalMicroarray genomics:KaryotypeAnalysis genomics:MLPA genomics:RepeatExpansionTest genomics:MethylationStudy genomics:RNASequencing ) ; sh:severity sh:Violation ; sh:name "Genetic Test Type"@en ; sh:message "GeneticTest must declare genomics:testType as one of the ten published genomics:TestType named individuals (SingleGeneTest, GenePanelTest, ExomeSequencing, GenomeSequencing, ChromosomalMicroarray, KaryotypeAnalysis, MLPA, RepeatExpansionTest, MethylationStudy, RNASequencing)"@en ] ; # WARNING: genePanel recommended — distinguishes 'no variant in BRCA1/2' # from 'no variant across a 47-gene panel'. sh:property [ sh:path genomics:genePanel ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Warning ; sh:name "Gene Panel"@en ; sh:message "genomics:genePanel is recommended; the analytical scope distinguishes a clinically meaningful negative result from absence-of-evidence"@en ] ; # WARNING: performingLab recommended for provenance. sh:property [ sh:path genomics:performingLab ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Warning ; sh:name "Performing Laboratory"@en ; sh:message "genomics:performingLab is recommended for provenance audit"@en ] . # ============================================================================ # Shape 10: Genetic Test Order # ============================================================================ # Note on naming: the task spec referenced 'orderDate', but v1-draft.0.1 names # the property genomics:orderedAt (xsd:dateTime). This shape uses orderedAt. genomics:GeneticTestOrderShape a sh:NodeShape ; sh:targetClass genomics:GeneticTestOrder ; rdfs:label "Genetic Test Order Shape"@en ; rdfs:comment "Validation constraints for genomics:GeneticTestOrder (FHIR ServiceRequest-aligned). Requires order status (closed enumeration over the four OrderStatus named individuals) and order timestamp (genomics:orderedAt)."@en ; # VIOLATION: orderStatus required and must be one of the four published # OrderStatus named individuals. sh:property [ sh:path genomics:orderStatus ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( genomics:OrderPending genomics:OrderInProgress genomics:OrderResulted genomics:OrderCancelled ) ; sh:severity sh:Violation ; sh:name "Order Status"@en ; sh:message "GeneticTestOrder must declare genomics:orderStatus as one of the four published genomics:OrderStatus named individuals (OrderPending, OrderInProgress, OrderResulted, OrderCancelled)"@en ] ; # VIOLATION: orderedAt required. sh:property [ sh:path genomics:orderedAt ; sh:datatype xsd:dateTime ; sh:minCount 1 ; sh:maxCount 1 ; sh:severity sh:Violation ; sh:name "Ordered At"@en ; sh:message "GeneticTestOrder must carry exactly one genomics:orderedAt xsd:dateTime"@en ] ; # WARNING: resultedIn — when status is OrderResulted, a fulfilled order # should link to its result. Pure SHACL cannot express the conditional # 'IF status=Resulted THEN resultedIn minCount 1' cleanly without sh:xone # patterns; we only constrain the range here. Conditional check lives in # the cascade-cli importer (TASK-1.x). sh:property [ sh:path genomics:resultedIn ; sh:class genomics:GeneticTest ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "Resulted In"@en ; sh:message "genomics:resultedIn, if present, must reference exactly one genomics:GeneticTest. When orderStatus is OrderResulted, this link should be populated (conditional check lives in cascade-cli importer)."@en ] . # ============================================================================ # Shape 11: Sequencing Run # ============================================================================ genomics:SequencingRunShape a sh:NodeShape ; sh:targetClass genomics:SequencingRun ; rdfs:label "Sequencing Run Shape"@en ; rdfs:comment "Validation constraints for genomics:SequencingRun. Coverage depth, sequencing technology, and reference genome are required — these three drive the data-quality tier inference upstream of the D-QUALITY-TIER safety constraint. Laboratory certification is optional but, when present, must be one of the four published values."@en ; # VIOLATION: coverageDepth required, decimal ≥ 0. sh:property [ sh:path genomics:coverageDepth ; sh:datatype xsd:decimal ; sh:minCount 1 ; sh:maxCount 1 ; sh:minInclusive 0.0 ; sh:severity sh:Violation ; sh:name "Coverage Depth"@en ; sh:message "SequencingRun must carry exactly one genomics:coverageDepth as a non-negative xsd:decimal (e.g., 30.0 for typical clinical WGS). Required for ClinicalGrade tier inference (>=30x WGS or validated panel-equivalent)."@en ] ; # VIOLATION: sequencingTechnology required and must be one of the five # published SequencingTechnology named individuals. sh:property [ sh:path genomics:sequencingTechnology ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( genomics:ShortReadIllumina genomics:LongReadONT genomics:LongReadPacBio genomics:Mixed genomics:GenotypingArray ) ; sh:severity sh:Violation ; sh:name "Sequencing Technology"@en ; sh:message "SequencingRun must declare genomics:sequencingTechnology as one of the five published genomics:SequencingTechnology named individuals (ShortReadIllumina, LongReadONT, LongReadPacBio, Mixed, GenotypingArray). GenotypingArray auto-tags ConsumerGrade in the importer."@en ] ; # VIOLATION: laboratoryCertification, when present, must be one of the # four published LaboratoryCertification named individuals. Optional # because some research/consumer runs legitimately have none. sh:property [ sh:path genomics:laboratoryCertification ; sh:maxCount 1 ; sh:in ( genomics:CLIA genomics:CAP genomics:ISO15189 genomics:Uncertified ) ; sh:severity sh:Violation ; sh:name "Laboratory Certification"@en ; sh:message "genomics:laboratoryCertification, if present, must be one of the four published genomics:LaboratoryCertification named individuals (CLIA, CAP, ISO15189, Uncertified)"@en ] ; # WARNING: variantCallerVersion recommended for reproducibility. sh:property [ sh:path genomics:variantCallerVersion ; sh:datatype xsd:string ; sh:maxCount 1 ; sh:severity sh:Warning ; sh:name "Variant Caller Version"@en ; sh:message "genomics:variantCallerVersion is recommended (e.g., 'GATK 4.5.0', 'DeepVariant 1.6'); needed to distinguish caller-driven differences from biological ones"@en ] ; # WARNING: sequencingDate recommended. sh:property [ sh:path genomics:sequencingDate ; sh:datatype xsd:date ; sh:maxCount 1 ; sh:severity sh:Warning ; sh:name "Sequencing Date"@en ] . # ============================================================================ # Shape 12: Raw File # ============================================================================ # Note: Cascade pods do NOT ingest raw genomic file bytes. RawFile carries # pointer-and-hash metadata only (see Critical conventions #9 in the # Genomics & Advisory implementation plan). genomics:RawFileShape a sh:NodeShape ; sh:targetClass genomics:RawFile ; rdfs:label "Raw File Shape"@en ; rdfs:comment "Validation constraints for genomics:RawFile pointer-and-hash records. Cascade does NOT ingest the bytes — only the manifest. Required: format (closed enum over seven FileFormat values), SHA-256 hash for integrity, opaque file location URI, and the reference genome the file is aligned against."@en ; # VIOLATION: fileFormat required and must be one of the seven published # FileFormat named individuals. sh:property [ sh:path genomics:fileFormat ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( genomics:BAM genomics:CRAM genomics:FASTQ genomics:gVCF genomics:VCF genomics:BCF genomics:OtherFileFormat ) ; sh:severity sh:Violation ; sh:name "File Format"@en ; sh:message "RawFile must declare genomics:fileFormat as one of the seven published genomics:FileFormat named individuals (BAM, CRAM, FASTQ, gVCF, VCF, BCF, OtherFileFormat). Hot tier (in-pod, encrypted): VCF/gVCF/BCF. Cold tier (out-of-pod, referenced): BAM/CRAM/FASTQ."@en ] ; # VIOLATION: fileHashSHA256 required — integrity check. sh:property [ sh:path genomics:fileHashSHA256 ; sh:datatype xsd:string ; sh:minCount 1 ; sh:maxCount 1 ; sh:minLength 64 ; sh:maxLength 64 ; sh:pattern "^[0-9a-f]{64}$" ; sh:severity sh:Violation ; sh:name "File SHA-256 Hash"@en ; sh:message "RawFile must carry exactly one genomics:fileHashSHA256 as a 64-character lowercase hex SHA-256 digest. Required for byte-on-disk integrity verification."@en ] ; # VIOLATION: fileLocation required (opaque URI; Cascade does NOT validate # or fetch this URI — only its presence and shape). sh:property [ sh:path genomics:fileLocation ; sh:datatype xsd:anyURI ; sh:minCount 1 ; sh:maxCount 1 ; sh:severity sh:Violation ; sh:name "File Location URI"@en ; sh:message "RawFile must carry exactly one genomics:fileLocation xsd:anyURI (opaque pointer to where the bytes live: s3://, https://, file://, vendor-specific). Cascade does NOT validate or fetch this URI."@en ] ; # VIOLATION: referenceGenome required. sh:property [ sh:path genomics:referenceGenome ; sh:datatype xsd:string ; sh:minCount 1 ; sh:maxCount 1 ; sh:minLength 1 ; sh:severity sh:Violation ; sh:name "Reference Genome"@en ; sh:message "RawFile must carry exactly one non-empty genomics:referenceGenome (e.g., 'GRCh38', 'GRCh37', 'T2T-CHM13'). Required for any downstream coordinate-based analysis."@en ] ; # VIOLATION: fileSizeBytes, when present, must be xsd:long ≥ 0. # xsd:long because BAM/CRAM/FASTQ files routinely exceed 2^31 bytes. sh:property [ sh:path genomics:fileSizeBytes ; sh:datatype xsd:long ; sh:maxCount 1 ; sh:minInclusive 0 ; sh:severity sh:Violation ; sh:name "File Size (Bytes)"@en ; sh:message "genomics:fileSizeBytes, if present, must be a non-negative xsd:long (BAM/CRAM/FASTQ files routinely exceed 2^31 bytes)"@en ] ; # INFO: htsgetEndpoint optional convenience. sh:property [ sh:path genomics:htsgetEndpoint ; sh:datatype xsd:anyURI ; sh:maxCount 1 ; sh:severity sh:Info ; sh:name "htsget Endpoint"@en ; sh:message "genomics:htsgetEndpoint is optional; when present, enables GA4GH htsget streaming of BAM/CRAM regions on demand"@en ] . # ============================================================================ # Cross-vocabulary note # ============================================================================ # - cascade:dataProvenance is declared in core/v1 and is not redeclared here. # genomics:dataProvenance is the genomics-specific origin lane (consumer- # array vs clinical-sequencing vs research-sequencing vs imported vs AI- # extracted) used by the importer to drive quality-tier auto-tagging. The # two are distinct and live in separate namespaces. # - checkup: terms (GeneticCounselingSummary, VariantNarrative) are out of # scope for this file and authored separately in the checkup vocabulary. # - prov:wasRevisionOf chain integrity for VariantInterpretation # reclassifications (the within-(variant, condition)-pair invariant) is a # cross-class invariant not expressible cleanly in pure SHACL Core; the # supplementing code-level check lives in cascade-cli (TASK-4.x advisory # applier validation). # ============================================================================ # Changelog # ============================================================================ # # Version 1.0-draft (2026-05-05) — TASK-0.2 # - Initial release of genomics vocabulary SHACL shapes. # - 12 sh:NodeShape covering Variant, VariantInterpretation, Haplotype, # Diplotype, CopyNumberVariant, SubmitterAssertion, Pedigree, # PedigreeMember, GeneticTest, GeneticTestOrder, SequencingRun, RawFile. # - D-Q5 enforced on VariantInterpretationShape: variantInterpreted and # condition each minCount/maxCount=1 (per-(variant, condition) modeling). # - D-QUALITY-TIER safety constraint enforced via sh:xone on # VariantInterpretationShape: classification non-pathogenic OR # (classification pathogenic-leaning AND mitigation present), where # mitigation = (variant.dataQualityTier=ClinicalGrade) OR # (requiresConfirmation=true). Violation message names both remediation # paths. # - Cross-class 'at least one stable identifier' on VariantShape via sh:or # over (caId, vrsId, clinvarVariationId, dbsnpRsId). # - Closed enumerations via sh:in for AcmgClass, OrderStatus, FileFormat, # SequencingTechnology, LaboratoryCertification, RelationshipRole, # CarrierStatus, ReviewStatus, TestType, DataQualityTier. # - Range / pattern checks for fileHashSHA256 (lowercase hex, 64 chars), # mosaicismFraction (decimal in [0.0, 1.0]), fileSizeBytes (xsd:long ≥ 0), # coverageDepth (decimal ≥ 0), CNV interval coordinates (positive int). # - Acknowledged vocabulary gaps (not enforced as SHACL violations): # * GeneticTest has no testDate property; chronology via # prov:generatedAtTime instead. # * SubmitterAssertion has no assertionDate property and no forward # reference to its parent Interpretation; the inverse link # (Interpretation→Assertion via genomics:aggregatedFrom) is # authoritative. # * Reclassification chain integrity (revising Interpretation MUST # share both variantInterpreted and condition with the revised one) # is a cross-focus-node invariant requiring sh:sparql; supplementing # code-level check lives in cascade-cli TASK-4.x. # * GeneticTestOrder.resultedIn conditional on orderStatus=Resulted is # a conditional cardinality not cleanly expressible in pure SHACL; # enforced by the importer. # - Out-of-scope (per TASK-0.2 scope): checkup: terms, advisory: cross- # vocab joins, LDPatch syntax, cryptographic verification. # PhenotypicObservationShape (v1-draft.0.4) genomics:PhenotypicObservationShape a sh:NodeShape ; sh:targetClass genomics:PhenotypicObservation ; sh:property [ sh:path genomics:hpoTerm ; sh:minCount 1 ; sh:datatype xsd:string ] ; sh:property [ sh:path genomics:phenotypePresence ; sh:minCount 1 ; sh:maxCount 1 ; sh:class genomics:PhenotypePresenceValue ] .