cs.DBSep 3, 2026
SaveA Semantic Model of Genetic Evidence: A Step Toward Bridging the Basic-Science-Clinic Gap
Organizations: Harvard University, Cambridge, MA, USA · IDEXX Laboratories, Westbrook, ME, USA · Forome Association, Newton, MA, USA · Deggendorf Institute of Technology, Germany
Abstract
Scientific and clinical decision-making depends on evidence from the primary literature, but existing standards for representing that evidence (FHIR Evidence, ECO, SEPIO, and the GA4GH Genomic Knowledge Standards) are oriented toward clinical-trial workflows, evidence codes, or single-variant assertions, and do not capture the fine-grained, domain-specific structure of claims in basic and pre-clinical research. We introduce a semantic model for scientific evidence with three core classes, specialize it for genetics, align it structurally to FHIR Evidence with a SEPIO-anchored credibility decomposition, and attach a compact dimensional vocabulary whose conditional-activation rules are validated by a SHACL schema for the implemented constraints. Using clinical variant interpretation as the driving use case, we evaluate the model through a human-AI annotation pilot over six genetics papers, yielding 28 evidence items and 95 source-anchored assertions, with a workflow that keeps curator-authored reference annotations distinct from AI-drafted annotations. Treating the pilot as a feasibility study rather than a benchmark, we argue that the model is a useful increment toward trustworthy, AI-ready infrastructure for variant interpretation: a reference data model and validation schema for representing genetic evidence.
Figures & tables
| This work | FHIR Evidence R5 | Relationship |
|---|---|---|
| ScientificEvidence | Evidence | Structural sibling; GEM replaces the PICO frame with a genetics-specific dimensional frame |
| EvidenceVariable | EvidenceVariable + Evidence.statistic | Broader; GEM combines the variable definition and measured value in one class |
| EvidenceAssertion | no equivalent | Meta-predicate over decision rules; part of the reasoning layer described in [ 4 ] , which is outside the scope of this paper |
| Credibility | Evidence.certainty | Structural parallel; both provide an overall rating with separable, genetics-specific subcomponents |
Table 1: Class-level correspondence between the core model and FHIR Evidence R5. The alignment is architectural: the genetics-specific dimensional frame replaces FHIR’s clinical-trial frame. Field-level detail is provided in Supplementary Table .
| Dimension | Value type | Cardinality | Objectivity | Example values |
|---|---|---|---|---|
| Knowledge Domain | categorical | multiple | obj. | human genetics , population genetics , model organism , gene function |
| Method | categorical | multiple | obj. | statistical genetics , in vivo , in vitro , clinical evidence |
| Target Type | categorical | single | obj. | gene , variant , interval , transcript |
| Resolution | categorical | single | obj. | window , gene , functional element , position , variant |
| Credibility | rating facets | multiple | subj. | overall rating sample_size , replication , multiple_testing , stratification_control , … |
| Phenotype Scale | categorical | single | obj. | molecular , cellular , histological , organismal , clinical |
Table 2: Always-required dimensions of the GeneticEvidence model. These apply to every evidence item regardless of context. Phenotype Scale was added to this set during the annotation work ( section 5 ). KD abbreviates Knowledge Domain; “obj.” and “subj.” denote objective and subjective, respectively.
| Dimension | Value type | Cardinality | Activation condition |
|---|---|---|---|
| Mode of Inheritance ‡ | categorical | single | KD human genetics AND TT = gene |
| Mendelian Segregation | boolean | single | KD human genetics AND TT = gene |
| Penetrance | categorical | single | KD human genetics AND TT = variant |
| Variant Ascertainment ‡ | categorical | multiple | TT = variant |
| Measurement Target | categorical | multiple | KD gene function |
| Gene Relation | categorical | single | KD gene function |
Table 3: Conditional dimensions of the GeneticEvidence model. Each is required only when its activation condition holds. KD, MT, and TT abbreviate Knowledge Domain, Method, and Target Type, respectively; denotes set membership for multi-valued dimensions. Dimensions marked ‡ have their activation condition enforced by an executable SHACL shape; for the others the condition is documented in the schema and checked at curation time, with executable shapes scoped as future work.
Figure 1: The two annotation tracks. Curator-authored annotations (left) form the feasibility corpus’s reference set. AI-drafted annotations (right) are produced under an annotation skill from the source PDF, the schema, and the manual annotations as exemplars, then reviewed by a curator under a review skill. Both tracks are validated against the same SHACL shapes. In every case, the curator, not the model, is the authority for the recorded annotation.
| Group | Papers | Count and examples |
|---|---|---|
| Finer structural resolution | J, N | 4 (subdomain, interaction, gene-relation, cross-level variant) |
| Polygenic-score machinery | IN | 6 (score target, whole-genome aggregate, target composition, epidemiological measurement, cohort sub-block, derived-artifact flag) |
| Effect-size and curation | DU, D | 3 (direction of effect, knowledge-domain priority, curator critique) |
| Cross-item and activation scope | DU | 2 (cross-item/translational synthesis, conditional-activation scope) |
Table 4: Unpromoted candidate schema extensions surfaced by the six-paper pilot, grouped by concern. Per-candidate detail is in the companion schema/EXTENSIONS.md ; paper keys are J = Jossin, D = Davis, N = Nelson, DU = Duerr, IN = Inouye. Gupta surfaced none.
| Claim | Where to verify |
|---|---|
| Canonical schema | schema/genetic_evidence.shacl.ttl (three enforced activation shapes: Variant Ascertainment, Mode of Inheritance, and Organism) |
| YAML-to-RDF transform | src/python/main/forome/gem/extraction/yaml_to_rdf.py (normalizes the v0/v1 field renames key_phrase / phrase and resolution / target_resolution ) |
| Corpus validation | gem-validate over annotations/ ; CI workflow .github/workflows/validate.yml |
| Coverage counts | gem-coverage regenerates annotations/coverage.md |
| Activation-shape tests | src/python/test/forome/gem/validation/ |
| UMLS crosswalk | data/umls/ (harness gem-umls-crosswalk ; decision log data/umls/DECISIONS.md ; SN8 regenerated by gem-umls-render ) |
Table 5: Artifact-to-claim traceability. Paths are relative to the repository root of the tagged release.
Appendix figures & tables16 assets
Supplementary material from the paper’s appendix.
Appendix
| Evidence pattern | FHIR | VA | SEPIO | ECO | ClinGen/ ACMG | GEM | Remaining gap (GEM) |
|---|---|---|---|---|---|---|---|
| 1. Mendelian pedigree: variant co-segregates with disease in a family | ◐ | ● | ◐ | ◐ | ● | ● | no structured segregation strength (meioses, LOD); activation scope for Variant targets (CE-DU4) |
| 2. Case–control burden: gene-level aggregate of rare variants | ◐ | ◐ | ◐ | ◐ | ● | ● | cohort composition (CE-IN5) |
| 3. Common-variant association: GWAS discovery, replication, fine mapping | ◐ | ◐ | ◐ | ◐ | ◐ | ● | direction of effect (CE-DU1); activation scope (CE-DU4); no discovery–replication link |
| 4. Functional assay: variant or gene effect measured in vitro | ◐ | ◐ | ◐ | ● | ● | ● | protein-subdomain resolution (CE-J1); assay type recorded as free text |
| 5. Animal model: knockout or conditional-knockout phenotype | ◐ | ◐ | ◐ | ● | ● | ● | HPO-only phenotype mapping; rescue not distinguished |
| 6. Gene–gene relation: X inhibits, binds, or regulates Y | ○ | ◐ | ◐ | ● | ● | ◐ | relation vocabulary (CE-J3); interaction as target (CE-J2) |
Table ST1: Ten evidence patterns across five standards and GEM. ● = represented natively; for ECO, a suitable term exists; for the combined ClinGen/ACMG column, the framework at the pattern’s level (gene or variant) provides a scored category or resulting classification. ◐ = represented only through an extension mechanism, only by reference to a resource outside the column’s module, or only in part; for ECO, only a generic term applies; for GEM, a forced fit is recorded as a candidate extension. ○ = not represented. – = outside the standard’s subject matter. The remaining gap column is relative to GEM and names a candidate extension where one exists. Row notes follow the table. FHIR = HL7 FHIR Evidence R5; VA = GA4GH Variant Annotation 1.0 with VRS 2.0 and Cat-VRS 1.0; ClinGen/ACMG = the ClinGen gene–disease validity framework, with other ClinGen products where noted, and the ACMG/AMP variant-interpretation guidelines.
| Paper | Drafted | Approved | With flags | Edited | Rejected | Added on review |
|---|---|---|---|---|---|---|
| Duerr 2006 | 6 | 0 | 2 | 4 | 0 | 1 (approved) |
| Inouye 2018 | 3 | 1 | 1 | 1 | 0 | 1 (edited) |
| Total | 9 | 1 | 3 | 5 | 0 | 2 |
Table ST2: Per-paper review verdicts for AI-drafted items. “With flags” denotes approval with non-blocking flags and no edits. “Added on review” denotes evidence omitted from the draft and entered by the reviewer as a new item.
| GEM construct | External term | Vocab. | Relation |
|---|---|---|---|
| ScientificEvidence (evidence item) | Evidence | FHIR R5 | structural sibling; GEM replaces the PICO frame with a genetics-specific dimensional frame |
| EvidenceVariable (definition and measured value) | EvidenceVariable + Evidence.statistic | FHIR R5 | broader; GEM combines the variable definition and measured value in one class |
| Credibility (overall rating slot) | Evidence.certainty | FHIR R5 | structural parallel; both provide an overall rating with separable subcomponents |
Table ST3: Structural correspondences between GEM and FHIR R5. The table extends the class-level alignment in the main paper with term-level detail.
| GEM construct | External term | Vocab. | Relation |
|---|---|---|---|
| assertion property (source-anchored claim) | Statement | GA4GH VA | narrower; a close, genetics-specific match |
| assertion property | Assertion ( SEPIO:0000001 ) | SEPIO | narrower |
| assertion property (false friend) | Evidence.assertion (human-readable summary) | FHIR R5 | none; do not map |
| EvidenceAssertion class (meta-predicate over a decision rule) | no equivalent | — | none |
Table ST4: Crosswalk for the assertion property and the EvidenceAssertion class. The property records an object-level, source-anchored claim; the class is a meta-predicate over a decision rule.
| GEM dimension / value | External term + ID | Vocab. | Relation |
|---|---|---|---|
| Target Type: gene | gene SO:0000704 | SO | exact |
| Target Type: related gene | gene SO:0000704 (relation recorded in Gene Relation) | SO | exact |
| Target Type: variant | sequence_variant SO:0001060 | SO | exact |
| Target Type: transcript | transcript SO:0000673 | SO | exact |
| Target Type: interval / segment | region SO:0000001 | SO | narrower |
| Mode of Inheritance (axis) | Mode of inheritance HP:0000005 | HPO | exact |
Table ST5: GEM dimensions and selected values mapped to external vocabularies. Identifiers were verified against their source vocabularies. SO = Sequence Ontology; VA = GA4GH Variant Annotation.
| GEM dimension | Why no external term |
|---|---|
| Knowledge Domain | no single external vocabulary represents this axis |
| Gene Relation (conditional) | GEM-specific gene–gene relations, such as X_regulates_Y and X_has_same_function_as_Y ; RO may cover some relations (unconfirmed) |
| Phenotype Scale (ordinal) | no external vocabulary provides this ordinal axis; GO biological_process is a possible future alignment for the molecular and cellular end of the scale |
| Variant Ascertainment | no single external term captures this dimension |
| Resolution | represented implicitly by the selected SO target-type term, rather than as an independent external axis |
| EvidenceAssertion (meta-predicate) | belongs to a reasoning layer beyond the scope of ECO, SEPIO, VA, and FHIR |
Table ST6: Coverage gaps in the present crosswalk. The relation is none in every row.
| ID | Claim | Target | Method | Cred. |
|---|---|---|---|---|
| GE-1 | rs11209026 protective in non-Jewish ileal CD (discovery) | variant | gwas | High |
| GE-2 | rs11209026 protective in Jewish ileal CD (replication) | variant | association study | High |
| GE-3 | Gln undertransmitted to affected offspring (FBAT) | variant | transmission disequilibrium test | High |
| GE-4 | Multiple independent IL23R signals (conditional) | variant | fine mapping | High |
| GE-5 | No association at IL12RB1/IL23A/IL12B | gene | gwas | High |
| GE-6 | Cited mouse-model and functional support | gene | in vivo , in vitro | High |
Table ST7: Duerr 2006: the seven GeneticEvidence items, with target type, leaf-level Method, and Credibility.
| ID | Claim | Target † | Cred. |
|---|---|---|---|
| GE-1 | Prediction and quintile stratification (HR 1.71/SD; 4.17-fold) | variant | High |
| GE-2 | Independence from six conventional risk factors | variant | High |
| GE-3 | Stratification persists among medicated individuals | variant | High |
| GE-new-1 | Combining three component scores outperforms each | variant | Medium |
Table ST8: Inouye 2018 ( v1 ): four GeneticEvidence items. Each uses the same placeholder Target Type and Resolution value, variant , and is flagged against CE-IN1 and CE-IN2. The actual target is a whole-genome polygenic score.
| Always-required dimensions | Conditional dimensions | ||||
|---|---|---|---|---|---|
| Dimension | Populated | Applicable | Dimension | Populated | Applicable |
| Knowledge Domain | 27 | 28 | Variant Ascertainment | 14 | 18 |
| Method | 28 | 28 | Mode of Inheritance | 0 | 1 |
| Target Type | 28 | 28 | Mendelian Segregation | 0 | 1 |
| Resolution | 28 | 28 | Penetrance | 2 | 4 |
| Credibility | 28 | 28 | Measurement Target | 10 | 11 |
Table ST9: Dimension population across 28 GeneticEvidence items from six papers (Duerr v0 , Inouye v1 ), computed by compute_coverage.py . The left block lists always-required dimensions; the right block lists conditional dimensions and the number of items in which their activation conditions apply.
| GEM value | CUI / TUI | Relation | UMLS concept or semantic type |
|---|---|---|---|
| knowledge_domain | |||
| (axis) | T090 | — | Occupation or Discipline (semantic type, tree A2.6) |
| HUMAN_GENETICS | C0020124 | exact | Human Genetics (Biomedical Occupation or Discipline; MSH) |
| ANIMAL_GENETICS | C1510895 | exact | Animal Genetics (Biomedical Occupation or Discipline; NCI) |
| POPULATION_GENETICS | C0017404 | exact | Genetics, Population (Occupation or Discipline; MSH) |
| COMPARATIVE_GENOMICS | C4704937 | exact | Comparative Genomics (Biomedical Occupation or Discipline; MSH) |
Table ST10: GEM dimensional vocabulary mapped to UMLS concepts (generated by the mapping harness and curator adjudication).
| GEM dimension | OMOP domain / table | Standard vocab. | Alignment and gap |
|---|---|---|---|
| phenotype_scale (CLINICAL) | Condition condition_occurrence | SNOMED CT | A clinical-scale phenotype aligns with an OMOP Condition concept (SNOMED CT), the CDM’s native representation of a clinical finding. Gap: Molecular, cellular, histological and organismal phenotype scales have no Condition-domain representation; OMOP has no phenotype-scale axis. |
| measurement_target (EXPRESSION, ACTIVITY, …) | Measurement measurement | LOINC | A quantitative molecular readout is conceptually a Measurement; LOINC covers some assay readouts (e.g. enzyme activity panels). Gap: Basic-science functional readouts (binding, localization, stability of a specific gene product in an experimental system) are rarely represented as standard LOINC Measurement concepts. |
| target_type (GENE) / resolution (GENE) | Observation / (Genomic CDM extension) observation (or OMOP Genomic CDM) | HGNC (via the OMOP Genomic vocabulary effort) | A gene identity aligns with HGNC; the OHDSI Genomic CDM extension is the locus where gene- and variant-level data are being standardized. Gap: The base OMOP CDM has no first-class gene or variant entity; this lives only in the evolving Genomic CDM extension, which is not yet a stable standard. |
| organism | (none) (none) | (none in CDM; NCBITaxon outside it) | No alignment. OMOP CDM is patient-centric (Homo sapiens implied) and has no concept of a model organism. Gap: Model-organism evidence is entirely out of scope for the CDM; GEM’s organism dimension maps to NCBITaxon ( Section SN5 ), not OMOP. |
| method (GWAS, ASSOCIATION_STUDY, …) | (metadata / study-level, not CDM clinical domains) (none; study design is not a CDM patient-data domain) | (none; partial coverage in MeSH/OBI outside OMOP) | No direct alignment. Study design / evidence-generation method is not a patient-level CDM domain. Gap: The CDM records observations, not how upstream genetic evidence was generated; GEM’s method hierarchy has no OMOP counterpart. |
Table ST11: Illustrative positioning of GEM dimensions against the OMOP CDM (hand-authored, not exhaustive).
| Concept family | Criterion failed | Examples | |
|---|---|---|---|
| “grade” (tumour, school, or product) | 202 | D | Astrocytoma, low grade (C1314694) |
| LOINC/PROMIS self-efficacy items | 191 | D | Current level of confidence I can drive a car (C5142440) |
| SNOMED degree qualifier values | 21 | A | Very high (C0442804); Extremely high (C5787630) |
| clinical “evidence of” expressions (presence of a sign) | 20 | D | Blood alcohol level of less than 20 mg/100 ml (C0481372); Blood alcohol level of 20–39 mg/100 ml (C0481373) |
| qualifiers denoting the degree of a measured property | 12 | A | Observation Value - High (C1561957); Message Waiting Priority - High (C1561958) |
| scores, probabilities, risk categories | 8 | A | IPSS-R Risk Category Very High (C5202916); IPSS Risk Category High (C4522209) |
Table ST12: Concepts retrieved by the credibility sweep (UMLS 2026AA), grouped by the adequacy criterion they failed. Criterion codes are defined in the text.
| GEM | GRADE certainty [ 8 ] | FHIR certainty-rating | CIO confidence level [ 3 ] | ACMG evidence strength [ 13 ] | ClinGen validity [ 12 ] |
|---|---|---|---|---|---|
| very_high | — (GRADE caps at high ) | high (lossy export) | high (partial) | very strong | definitive |
| high | high | high | high | strong | strong |
| medium | moderate | moderate | medium | moderate | moderate |
| low | low | low | low (“absence of trust”, closer to none) | supporting | limited |
| — | very low | very-low | — | — | — |
Table ST13: Conceptual, non-positional alignment of the GEM credibility tiers with selected external scales. “—” indicates that no counterpart is proposed. The FHIR column gives the lossy export used when a certainty-rating code is required [ 9 ] . CIO = Confidence Information Ontology [ 3 ] .
| Paper | Source | GE items | Credibility | Role | Notes |
|---|---|---|---|---|---|
| Jossin 2017 (Llgl1) | manual | 6 | Very high | molecular mechanism | multi-scale phenotype; motivates phenotype_scale |
| Davis 2011 (TTC21B) | manual | 6 | High | breadth exemplar | motivates variant_ascertainment |
| Nelson 1992 (CD18) | manual | 3 | Medium | classical molecular genetics | single-gene functional; one candidate (CE-N1) resolved by external reference |
| Gupta 2015 (ATP6AP2) | manual | 2 | Low | low-credibility edge case | schema fits cleanly |
| Duerr 2006 (IL23R) | AI-drafted | 7 | High | clean GWAS exemplar | featured in the main text ( v0 ) |
| Inouye 2018 (metaGRS) | AI-drafted | 4 | High/Med | model-extension stress test | featured in the main text ( v1 ) |
Table ST14: The six annotated publications and the epistemic shape each exercises. Manual annotations form the curator-authored reference set; AI-drafted annotations are evaluated against curator review. GE abbreviates GeneticEvidence ; the “GE items” column counts how many evidence items each paper yields under our claim-level decomposition.
| Facet | Aspect of methodological soundness recorded |
|---|---|
| sample_size | Size of cohort, pedigree, or sample (statistical power) |
| ascertainment | Case/control makeup and selection of the studied sample |
| multiple_testing | Whether and how multiplicity was controlled |
| stratification_control | Ancestry restriction or matching, and residual-confounding control |
| replication | Independent replication or internal resampling |
Table ST15: Credibility facets in the GeneticEvidence model. Credibility records methodological soundness only; each facet is rated separately rather than collapsed to one score, and the facets are registered as genetics-specific subcomponents anchored on SEPIO confidence. The overall rating’s levels are analogous in spirit to ClinGen gene-disease validity classifications [ 12 ] , which however aggregate per gene-disease relationship rather than per assertion. Effect magnitude is not a credibility facet: it is a measured value carried on the GeneticEvidenceVariable . Phenotype fit to a clinical question is applicability, captured by Relevance and Specificity.
| This work | FHIR Evidence R5 | Note |
|---|---|---|
| ScientificEvidence item | Evidence | One evidence concept per claim |
| dimensions | Evidence.variableDefinition (roles) | PICO roles replaced by genetic dimensions |
| EvidenceVariable definition | EvidenceVariable.characteristic | What the variable is and how measured |
| EvidenceVariable value | Evidence.statistic | The measured quantity itself |
| Credibility overall rating | Evidence.certainty.rating | Coded ordinal rating anchored on SEPIO confidence |
| Credibility facet | Evidence.certainty.certaintySubcomponent.type | Extension to the subcomponent-type value set |
Table ST16: Field-level alignment of the core model to FHIR Evidence R5. The certainty rows expand the Credibility entry of the class-level alignment table in the main paper.
More in this topic
Knowledge Representation
Knowledge representation studies how to encode facts, concepts, rules, context, and uncertainty so machines can interpret and reason over them.
See all 128 papers