Skip to content

Class: ModelMechanismLink

Links an experimental (NAM), animal, or computational model to a specific pathophysiology mechanism node, recording which facet of that mechanism the model recapitulates, perturbs, or reads out; how faithfully it does so; and the outcome measures that ground the claim.

URI: dismech:class/ModelMechanismLink

 classDiagram
    class ModelMechanismLink
    click ModelMechanismLink href "../../classes/ModelMechanismLink/"
      ModelMechanismLink : description

      ModelMechanismLink : divergences





        ModelMechanismLink --> "*" ModelDivergence : divergences
        click ModelDivergence href "../../classes/ModelDivergence/"



      ModelMechanismLink : evidence





        ModelMechanismLink --> "* _recommended_" EvidenceItem : evidence
        click EvidenceItem href "../../classes/EvidenceItem/"



      ModelMechanismLink : fidelity





        ModelMechanismLink --> "0..1" ModelFidelityEnum : fidelity
        click ModelFidelityEnum href "../../enums/ModelFidelityEnum/"



      ModelMechanismLink : limitations

      ModelMechanismLink : model_scale





        ModelMechanismLink --> "0..1" BiologicalScaleEnum : model_scale
        click BiologicalScaleEnum href "../../enums/BiologicalScaleEnum/"



      ModelMechanismLink : readouts





        ModelMechanismLink --> "*" ExperimentalReadout : readouts
        click ExperimentalReadout href "../../classes/ExperimentalReadout/"



      ModelMechanismLink : relationship





        ModelMechanismLink --> "0..1" ModelMechanismRelationshipEnum : relationship
        click ModelMechanismRelationshipEnum href "../../enums/ModelMechanismRelationshipEnum/"



      ModelMechanismLink : target

Slots

Name Cardinality and Range Description Inheritance
target 1
String		 Name of the pathophysiology entry this model is linked to direct
relationship 0..1
ModelMechanismRelationshipEnum		 What this model does to the linked mechanism: reproduces it, manipulates it, ... direct
description 0..1
String		 Brief assertion-level note describing what facet of the linked mechanism the ... direct
readouts *
ExperimentalReadout		 Outcome measures through which this model reports on the linked mechanism, ea... direct
fidelity 0..1
ModelFidelityEnum		 Curator assessment of how faithfully this model captures the linked human mec... direct
limitations 0..1
String		 Specific caveats on transferring findings from this model to the human mechan... direct
model_scale 0..1
BiologicalScaleEnum		 Biological scale at which this model actually represents or measures the link... direct
divergences *
ModelDivergence		 Typed, individually explained ways in which this model departs from the mecha... direct
evidence * recommended
EvidenceItem		 Evidence that this model is informative for the linked mechanism direct

Usages

used by used in type used
ExperimentalModel modeled_mechanisms range ModelMechanismLink
ComputationalModel modeled_mechanisms range ModelMechanismLink
AnimalModel modeled_mechanisms range ModelMechanismLink

Comments

  • Two evidence layers, deliberately separate. Evidence on the link attests that the model is informative for the node; evidence on each readout attests that a specific measurement was made, in a specific direction.
  • Readouts belong on the link rather than on the model, because one model typically measures different things for different mechanism nodes.
  • A FAILS_TO_RECAPITULATE link is a substantive negative claim and should carry both limitations and evidence; it usually pairs with a HUMAN_MODEL_MISMATCH discussion.
  • A link is a context-of-use statement in the sense of the model-credibility literature: it asserts that this model is informative for this specific mechanism, which is a narrower and more checkable claim than "this model is good". divergences is the corresponding applicability analysis. The framing is borrowed from ASME V&V 40 and the FDA credibility guidance (both written for medical-device submissions) and from the healthcare-wide ten rules of credible practice, whose rule 1 is "define context clearly" and rule 4 "list limitations explicitly".
  • model_scale records the scale the model observes, against which the target node's own biological_scale can be compared. The comparison is directional and the two directions mean different things: a model BELOW its target's scale is extrapolating upward — it cannot observe the outcome it is cited for — whereas a model ABOVE its target's scale contains that scale and is normally unremarkable (a whole animal can report a molecular readout). Do not store the comparison; derive it.

See Also

Identifier and Mapping Information

Schema Source

  • from schema: https://w3id.org/monarch-initiative/dismech

Mappings

Mapping Type Mapped Value
self dismech:ModelMechanismLink
native dismech:ModelMechanismLink

LinkML Source

Direct

name: ModelMechanismLink
description: Links an experimental (NAM), animal, or computational model to a specific
  pathophysiology mechanism node, recording which facet of that mechanism the model
  recapitulates, perturbs, or reads out; how faithfully it does so; and the outcome
  measures that ground the claim.
comments:
- Two evidence layers, deliberately separate. Evidence on the link attests that the
  model is informative for the node; evidence on each readout attests that a specific
  measurement was made, in a specific direction.
- Readouts belong on the link rather than on the model, because one model typically
  measures different things for different mechanism nodes.
- A FAILS_TO_RECAPITULATE link is a substantive negative claim and should carry both
  `limitations` and `evidence`; it usually pairs with a HUMAN_MODEL_MISMATCH discussion.
- 'A link is a *context-of-use* statement in the sense of the model-credibility literature:
  it asserts that this model is informative for this specific mechanism, which is
  a narrower and more checkable claim than "this model is good". `divergences` is
  the corresponding applicability analysis. The framing is borrowed from ASME V&V
  40 and the FDA credibility guidance (both written for medical-device submissions)
  and from the healthcare-wide ten rules of credible practice, whose rule 1 is "define
  context clearly" and rule 4 "list limitations explicitly".'
- '`model_scale` records the scale the model observes, against which the target node''s
  own `biological_scale` can be compared. The comparison is directional and the two
  directions mean different things: a model BELOW its target''s scale is extrapolating
  upward — it cannot observe the outcome it is cited for — whereas a model ABOVE its
  target''s scale contains that scale and is normally unremarkable (a whole animal
  can report a molecular readout). Do not store the comparison; derive it.'
from_schema: https://w3id.org/monarch-initiative/dismech
see_also:
- PMID:32993675
- https://www.asme.org/codes-standards/find-codes-standards/assessing-credibility-of-computational-modeling-through-verification-and-validation-application-to-medical-devices
- https://www.fda.gov/regulatory-information/search-fda-guidance-documents/assessing-credibility-computational-modeling-and-simulation-medical-device-submissions
slots:
- target
- relationship
- description
- readouts
- fidelity
- limitations
- model_scale
- divergences
- evidence
slot_usage:
  target:
    name: target
    description: Name of the pathophysiology entry this model is linked to. Must match
      a pathophysiology name in the same disease file.
  relationship:
    name: relationship
    description: 'What this model does to the linked mechanism: reproduces it, manipulates
      it, measures it, rescues it, or fails to reproduce it.'
    range: ModelMechanismRelationshipEnum
  description:
    name: description
    description: Brief assertion-level note describing what facet of the linked mechanism
      the model captures or assays.
  readouts:
    name: readouts
    description: Outcome measures through which this model reports on the linked mechanism,
      each optionally grounded to an OBI assay, an HP phenotype, a biomarker, or a
      GO process, with a measured direction and its own evidence. Each readout's own
      `target` is required and must repeat the link's `target`, so a readout stays
      self-describing when lifted out of its link by the graph and KGX exporters.
    range: ExperimentalReadout
    inlined_as_list: true
  evidence:
    name: evidence
    description: Evidence that this model is informative for the linked mechanism
    recommended: true

Induced

name: ModelMechanismLink
description: Links an experimental (NAM), animal, or computational model to a specific
  pathophysiology mechanism node, recording which facet of that mechanism the model
  recapitulates, perturbs, or reads out; how faithfully it does so; and the outcome
  measures that ground the claim.
comments:
- Two evidence layers, deliberately separate. Evidence on the link attests that the
  model is informative for the node; evidence on each readout attests that a specific
  measurement was made, in a specific direction.
- Readouts belong on the link rather than on the model, because one model typically
  measures different things for different mechanism nodes.
- A FAILS_TO_RECAPITULATE link is a substantive negative claim and should carry both
  `limitations` and `evidence`; it usually pairs with a HUMAN_MODEL_MISMATCH discussion.
- 'A link is a *context-of-use* statement in the sense of the model-credibility literature:
  it asserts that this model is informative for this specific mechanism, which is
  a narrower and more checkable claim than "this model is good". `divergences` is
  the corresponding applicability analysis. The framing is borrowed from ASME V&V
  40 and the FDA credibility guidance (both written for medical-device submissions)
  and from the healthcare-wide ten rules of credible practice, whose rule 1 is "define
  context clearly" and rule 4 "list limitations explicitly".'
- '`model_scale` records the scale the model observes, against which the target node''s
  own `biological_scale` can be compared. The comparison is directional and the two
  directions mean different things: a model BELOW its target''s scale is extrapolating
  upward — it cannot observe the outcome it is cited for — whereas a model ABOVE its
  target''s scale contains that scale and is normally unremarkable (a whole animal
  can report a molecular readout). Do not store the comparison; derive it.'
from_schema: https://w3id.org/monarch-initiative/dismech
see_also:
- PMID:32993675
- https://www.asme.org/codes-standards/find-codes-standards/assessing-credibility-of-computational-modeling-through-verification-and-validation-application-to-medical-devices
- https://www.fda.gov/regulatory-information/search-fda-guidance-documents/assessing-credibility-computational-modeling-and-simulation-medical-device-submissions
slot_usage:
  target:
    name: target
    description: Name of the pathophysiology entry this model is linked to. Must match
      a pathophysiology name in the same disease file.
  relationship:
    name: relationship
    description: 'What this model does to the linked mechanism: reproduces it, manipulates
      it, measures it, rescues it, or fails to reproduce it.'
    range: ModelMechanismRelationshipEnum
  description:
    name: description
    description: Brief assertion-level note describing what facet of the linked mechanism
      the model captures or assays.
  readouts:
    name: readouts
    description: Outcome measures through which this model reports on the linked mechanism,
      each optionally grounded to an OBI assay, an HP phenotype, a biomarker, or a
      GO process, with a measured direction and its own evidence. Each readout's own
      `target` is required and must repeat the link's `target`, so a readout stays
      self-describing when lifted out of its link by the graph and KGX exporters.
    range: ExperimentalReadout
    inlined_as_list: true
  evidence:
    name: evidence
    description: Evidence that this model is informative for the linked mechanism
    recommended: true
attributes:
  target:
    name: target
    description: Name of the pathophysiology entry this model is linked to. Must match
      a pathophysiology name in the same disease file.
    from_schema: https://w3id.org/monarch-initiative/dismech
    rank: 1000
    alias: target
    owner: ModelMechanismLink
    domain_of:
    - ExperimentalPerturbation
    - ExperimentalReadout
    - CausalEdge
    - TreatmentMechanismTarget
    - EnvironmentalMechanismTarget
    - ModelMechanismLink
    - BiomarkerReadout
    - PhenotypeReadout
    range: string
    required: true
  relationship:
    name: relationship
    description: 'What this model does to the linked mechanism: reproduces it, manipulates
      it, measures it, rescues it, or fails to reproduce it.'
    from_schema: https://w3id.org/monarch-initiative/dismech
    rank: 1000
    alias: relationship
    owner: ModelMechanismLink
    domain_of:
    - GeneSetAssociation
    - ModelMechanismLink
    - BiomarkerReadout
    - PhenotypeReadout
    range: ModelMechanismRelationshipEnum
  description:
    name: description
    description: Brief assertion-level note describing what facet of the linked mechanism
      the model captures or assays.
    from_schema: https://w3id.org/monarch-initiative/dismech
    rank: 1000
    alias: description
    owner: ModelMechanismLink
    domain_of:
    - Descriptor
    - DietaryModification
    - GeneticContext
    - Dataset
    - ExperimentalModel
    - Experiment
    - ExperimentalPerturbation
    - ExperimentalReadout
    - ExperimentalControl
    - ClinicalTrial
    - ComputationalModel
    - ModelVariable
    - DifferentialDiagnosis
    - Subtype
    - CausalEdge
    - TreatmentMechanismTarget
    - EnvironmentalMechanismTarget
    - ModelDivergence
    - ModelMechanismLink
    - BiomarkerReadout
    - PhenotypeReadout
    - SurrogateEndpointCollection
    - ProteinStructure
    - ExternalAssertion
    - EpidemiologyInfo
    - Pathophysiology
    - Phenotype
    - HistopathologyFinding
    - ImagingFinding
    - Environmental
    - Disease
    - Stage
    - AgentLifeCycle
    - AgentLifeCycleStage
    - AnimalModel
    - Treatment
    - InfectiousAgent
    - Transmission
    - Assay
    - Diagnosis
    - Inheritance
    - Variant
    - FunctionalEffect
    - Mechanism
    - ModelingConsideration
    - Definition
    - CriteriaSet
    - ConditionDescriptor
    - GOEnrichment
    - ComorbidityHypothesis
    - UpstreamConditionHypothesis
    - MechanisticHypothesis
    - Grouping
    - GroupingCriteria
    - LogicalCriterion
    - DifferentiatingMechanism
    - ModuleCollection
    - ModuleCollectionMember
    range: string
  readouts:
    name: readouts
    description: Outcome measures through which this model reports on the linked mechanism,
      each optionally grounded to an OBI assay, an HP phenotype, a biomarker, or a
      GO process, with a measured direction and its own evidence. Each readout's own
      `target` is required and must repeat the link's `target`, so a readout stays
      self-describing when lifted out of its link by the graph and KGX exporters.
    comments:
    - Target names should match pathophysiology or phenotype entry names in the same
      disease file
    - Readout links are observational/associative, not causal disease-progression
      edges
    - Use evidence on the readout link when the biomarker-to-mechanism mapping is
      distinct from the biomarker's own evidence
    from_schema: https://w3id.org/monarch-initiative/dismech
    rank: 1000
    alias: readouts
    owner: ModelMechanismLink
    domain_of:
    - Experiment
    - ModelMechanismLink
    - Biochemical
    range: ExperimentalReadout
    multivalued: true
    inlined: true
    inlined_as_list: true
  fidelity:
    name: fidelity
    description: Curator assessment of how faithfully this model captures the linked
      human mechanism. A translational-validity caveat, not a metric.
    from_schema: https://w3id.org/monarch-initiative/dismech
    rank: 1000
    alias: fidelity
    owner: ModelMechanismLink
    domain_of:
    - ModelMechanismLink
    range: ModelFidelityEnum
  limitations:
    name: limitations
    description: 'Specific caveats on transferring findings from this model to the
      human mechanism: species divergence, supraphysiological expression, absent cell
      types, missing immune or vascular compartments, and similar.'
    from_schema: https://w3id.org/monarch-initiative/dismech
    rank: 1000
    alias: limitations
    owner: ModelMechanismLink
    domain_of:
    - ModelMechanismLink
    range: string
  model_scale:
    name: model_scale
    description: 'Biological scale at which this model actually represents or measures
      the linked target — molecular, cellular, tissue/organ, or organism. Records
      what the model *observes*, which is not always the scale of the mechanism it
      is cited for: a signalling-network model whose output node is named for a tissue-level
      outcome still observes only molecular state. Compared against the target node''s
      own `biological_scale`, this makes the scale gap computable rather than leaving
      it buried in `limitations` prose. Reuses BiologicalScaleEnum so the two are
      directly comparable.'
    examples:
    - value: MOLECULAR
      description: A Boolean signalling network linked to an ORGANISM-scale node —
        the model observes molecular state and infers the organism-scale outcome.
    - value: ORGANISM
      description: A whole-animal model linked to a MOLECULAR node — the model contains
        the target scale, so no extrapolation is involved.
    from_schema: https://w3id.org/monarch-initiative/dismech
    rank: 1000
    alias: model_scale
    owner: ModelMechanismLink
    domain_of:
    - ModelMechanismLink
    range: BiologicalScaleEnum
  divergences:
    name: divergences
    description: 'Typed, individually explained ways in which this model departs from
      the mechanism it is linked to. The structured counterpart of the free-text `limitations`
      summary: each entry names a kind from a fixed taxonomy, says in the curator''s
      own words why that kind of gap applies *here*, and optionally records whether
      it bears on this link''s claim. Multivalued because a real caveat is usually
      several kinds at once.'
    comments:
    - Complements `relationship` and `fidelity` rather than replacing them -- those
      say what the model does and how well, this says in what specific respects it
      falls short
    - A SCALE_EXTRAPOLATION divergence should agree with the `model_scale` comparison;
      `just model-scale-audit` cross-checks the two
    - This slot is the structured form of rule 4 of the ten rules of credible practice
      in healthcare modeling ("list limitations explicitly"); typing the list rather
      than writing it as prose is the only change dismech makes to that rule.
    from_schema: https://w3id.org/monarch-initiative/dismech
    see_also:
    - PMID:32993675
    rank: 1000
    alias: divergences
    owner: ModelMechanismLink
    domain_of:
    - ModelMechanismLink
    range: ModelDivergence
    multivalued: true
    inlined: true
    inlined_as_list: true
  evidence:
    name: evidence
    description: Evidence that this model is informative for the linked mechanism
    from_schema: https://w3id.org/monarch-initiative/dismech
    rank: 1000
    alias: evidence
    owner: ModelMechanismLink
    domain_of:
    - PhenotypeContext
    - Dataset
    - ExperimentalModel
    - Experiment
    - ExperimentalPerturbation
    - ExperimentalReadout
    - ExperimentalControl
    - ClinicalTrial
    - ComputationalModel
    - DifferentialDiagnosis
    - Subtype
    - CausalEdge
    - TreatmentMechanismTarget
    - EnvironmentalMechanismTarget
    - ModelDivergence
    - ModelMechanismLink
    - BiomarkerReadout
    - PhenotypeReadout
    - ReferenceRange
    - SurrogateEndpoint
    - ExternalAssertion
    - Finding
    - Prevalence
    - GeneCaseFraction
    - ProgressionInfo
    - ClinicalBurden
    - EpidemiologyInfo
    - Pathophysiology
    - Phenotype
    - Biochemical
    - HistopathologyFinding
    - ImagingFinding
    - Genetic
    - Environmental
    - Stage
    - AgentLifeCycle
    - AgentLifeCycleStage
    - AnimalModel
    - Treatment
    - InfectiousAgent
    - Transmission
    - Diagnosis
    - Inheritance
    - Variant
    - ModelingConsideration
    - ClassificationAssignment
    - Definition
    - AlgorithmValidationStatus
    - CriteriaSet
    - AssociationSignal
    - AssociationStatistics
    - ComorbidityHypothesis
    - UpstreamConditionHypothesis
    - MechanisticHypothesis
    - Discussion
    - GroupingCriteria
    - GroupingMember
    - DifferentiatingMechanism
    - ModuleCollection
    - ModuleCollectionMember
    range: EvidenceItem
    recommended: true
    multivalued: true
    inlined: true
    inlined_as_list: true