Skip to content

Class: LatentDecompositionDetails

Payload for an analysis whose map is one component or latent variable of a decomposition -- ICA, PCA, NMF, PLS, CCA, and the multimodal fusion methods. Covers a component whose map or loading was tested: compared between cohorts, correlated with behaviour, or otherwise given an inferential result. A component map presented descriptively with no test is not an Analysis, so it takes no record here. Being representable here does not imply eligibility for coordinate-based meta-analysis.

One analysis reports one component: a second latent variable or component from the same decomposition is a second Analysis.

One invariant LinkML does not express, so the structural validator carries it: crossblock_covariance_percent is meaningless when second_block is [none], there being no second block for the covariance to cross. variance_explained_percent is permitted either way, since a two-block decomposition may still report how much of the brain block its component captures.

URI: neurostudy:LatentDecompositionDetails

 classDiagram
    class LatentDecompositionDetails
    click LatentDecompositionDetails href "../LatentDecompositionDetails/"
      AnalysisDetails <|-- LatentDecompositionDetails
        click AnalysisDetails href "../AnalysisDetails/"

      LatentDecompositionDetails : component_id

      LatentDecompositionDetails : component_label

      LatentDecompositionDetails : crossblock_covariance_percent

      LatentDecompositionDetails : details_type

      LatentDecompositionDetails : dimensionality_selection_method

      LatentDecompositionDetails : map_quantity

      LatentDecompositionDetails : method

      LatentDecompositionDetails : polarity_semantics





        LatentDecompositionDetails --> "0..1" PolaritySemantics : polarity_semantics
        click PolaritySemantics href "../PolaritySemantics/"



      LatentDecompositionDetails : salience_threshold

      LatentDecompositionDetails : second_block

      LatentDecompositionDetails : second_block_assessments





        LatentDecompositionDetails --> "*" Assessment : second_block_assessments
        click Assessment href "../Assessment/"



      LatentDecompositionDetails : second_block_detail

      LatentDecompositionDetails : selection_criterion

      LatentDecompositionDetails : total_components

      LatentDecompositionDetails : variance_explained_percent

Inheritance

Slots

Name Cardinality and Range Description Inheritance
method 1
String or 
DecompositionMethod		 The algorithm that produced the components, which decides what their values a... direct
second_block 1..*
String or 
SecondBlock		 What the brain block was decomposed against, one value per additional block, ... direct
second_block_assessments *
Assessment		 The measures forming a behavioural or clinical block, by identifier, when the... direct
second_block_detail *
String		 What each non-assessment block held, in the source's words: the seed regions ... direct
component_id 1
String		 Identifier of the component or latent variable within the decomposition, as r... direct
component_label 0..1
String		 The source's name for the component, such as a network label, or for what a l... direct
total_components 0..1
Integer		 Number of components or latent variables the decomposition produced direct
dimensionality_selection_method 0..1
String		 How the number of components was chosen, as reported -- minimum description l... direct
selection_criterion 0..1
String		 How this component was selected for reporting direct
variance_explained_percent 0..1
Float		 Percentage of the brain block's variance this component accounts for, as repo... direct
crossblock_covariance_percent 0..1
Float		 Percentage of crossblock covariance this latent variable accounts for, as rep... direct
map_quantity 0..1
String or 
MapQuantity		 What the values of the reported map represent -- a loading, a weight, a PLS s... direct
salience_threshold 0..1
String		 The criterion deciding which elements count as reliable contributors, as repo... direct
polarity_semantics 0..1
PolaritySemantics		 Whether the sign of the reported map carries meaning direct
details_type 1
String		 The concrete details subclass carried by this payload AnalysisDetails

Identifier and Mapping Information

Schema Source

  • from schema: https://neurostuff.org/schemas/neuroimaging-study-storage

Mappings

Mapping Type Mapped Value
self neurostudy:LatentDecompositionDetails
native neurostudy:LatentDecompositionDetails

LinkML Source

Direct

name: LatentDecompositionDetails
description: 'Payload for an analysis whose map is one component or latent variable
  of a decomposition -- ICA, PCA, NMF, PLS, CCA, and the multimodal fusion methods.
  Covers a component whose map or loading was *tested*: compared between cohorts,
  correlated with behaviour, or otherwise given an inferential result. A component
  map presented descriptively with no test is not an Analysis, so it takes no record
  here. Being representable here does not imply eligibility for coordinate-based meta-analysis.

  One analysis reports one component: a second latent variable or component from the
  same decomposition is a second Analysis.

  One invariant LinkML does not express, so the structural validator carries it: `crossblock_covariance_percent`
  is meaningless when `second_block` is `[none]`, there being no second block for
  the covariance to cross. `variance_explained_percent` is permitted either way, since
  a two-block decomposition may still report how much of the brain block its component
  captures.'
from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage
rank: 1000
is_a: AnalysisDetails
attributes:
  method:
    name: method
    description: The algorithm that produced the components, which decides what their
      values and signs mean.
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    required: true
    any_of:
    - range: DecompositionMethod
    - range: string
  second_block:
    name: second_block
    description: 'What the brain block was decomposed against, one value per additional
      block, and the primary filter on this payload. `[none]` for a decomposition
      of the brain data alone, which is a positive assertion rather than an empty
      slot.

      Multivalued because papers cross more than two blocks: a three-modality mCCA+jICA
      carries `other_imaging_modality` twice, and a design-plus-behaviour PLS carries
      `[design, behavioral]`. Name the blocks, never their number.'
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    required: true
    multivalued: true
    any_of:
    - range: SecondBlock
    - range: string
  second_block_assessments:
    name: second_block_assessments
    description: The measures forming a behavioural or clinical block, by identifier,
      when they are `Assessment` records of this study. Empty when the second block
      is the design, another imaging modality, or anything else that is not an assessment
      -- those are named in `second_block_detail`.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: Assessment
    multivalued: true
    inlined: false
  second_block_detail:
    name: second_block_detail
    description: 'What each non-assessment block held, in the source''s words: the
      seed regions of a seed decomposition, the modalities of a fusion ("ALFF, FA,
      grey matter"), the genotype array of a parallel ICA ("367 SNPs"). Free text
      because the range of things that form a second block is open, and a SNP array
      is neither an `Assessment` nor an `Acquisition` of this study.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: string
    multivalued: true
  component_id:
    name: component_id
    description: Identifier of the component or latent variable within the decomposition,
      as reported -- commonly an index such as `IC12` or `LV1`. Required because without
      it two analyses from one decomposition are indistinguishable.
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: string
    required: true
  component_label:
    name: component_label
    description: The source's name for the component, such as a network label, or
      for what a latent variable captures.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: string
  total_components:
    name: total_components
    description: Number of components or latent variables the decomposition produced.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: integer
    minimum_value: 0
  dimensionality_selection_method:
    name: dimensionality_selection_method
    description: How the number of components was chosen, as reported -- minimum description
      length, a variance-retained cutoff, a scree criterion, permutation on the singular
      values, or a fixed a priori number.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: string
  selection_criterion:
    name: selection_criterion
    description: How this component was selected for reporting.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: string
  variance_explained_percent:
    name: variance_explained_percent
    description: 'Percentage of the brain block''s variance this component accounts
      for, as reported. Not interchangeable with `crossblock_covariance_percent`:
      one is variance within a block and the other covariance between blocks, and
      averaging across the two would combine incomparable quantities.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: float
    minimum_value: 0
    maximum_value: 100
  crossblock_covariance_percent:
    name: crossblock_covariance_percent
    description: Percentage of crossblock covariance this latent variable accounts
      for, as reported. The nearest thing a two-block decomposition has to an effect
      size, and what a paper uses to justify reporting one latent variable and not
      another. Unset when `second_block` is `[none]`, there being no second block
      to cross.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: float
    minimum_value: 0
    maximum_value: 100
  map_quantity:
    name: map_quantity
    description: What the values of the reported map represent -- a loading, a weight,
      a PLS salience, a standardized score.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    any_of:
    - range: MapQuantity
    - range: string
  salience_threshold:
    name: salience_threshold
    description: 'The criterion deciding which elements count as reliable contributors,
      as reported -- conventionally a bootstrap ratio cutoff such as "|BSR| > 3.0",
      or a standardized cutoff such as "|Z| > 3.5". Not `InferenceSettings`: a bootstrap
      ratio is a stability measure and not a test statistic, so reading it as a voxelwise
      threshold would misstate the inference.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: string
  polarity_semantics:
    name: polarity_semantics
    description: Whether the sign of the reported map carries meaning. A latent variable's
      sign is arbitrary up to a flip, and so is an independent component's, so a paper
      must say what positive means before a direction can be pooled.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    domain_of:
    - LatentDecompositionDetails
    range: PolaritySemantics

Induced

name: LatentDecompositionDetails
description: 'Payload for an analysis whose map is one component or latent variable
  of a decomposition -- ICA, PCA, NMF, PLS, CCA, and the multimodal fusion methods.
  Covers a component whose map or loading was *tested*: compared between cohorts,
  correlated with behaviour, or otherwise given an inferential result. A component
  map presented descriptively with no test is not an Analysis, so it takes no record
  here. Being representable here does not imply eligibility for coordinate-based meta-analysis.

  One analysis reports one component: a second latent variable or component from the
  same decomposition is a second Analysis.

  One invariant LinkML does not express, so the structural validator carries it: `crossblock_covariance_percent`
  is meaningless when `second_block` is `[none]`, there being no second block for
  the covariance to cross. `variance_explained_percent` is permitted either way, since
  a two-block decomposition may still report how much of the brain block its component
  captures.'
from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage
rank: 1000
is_a: AnalysisDetails
attributes:
  method:
    name: method
    description: The algorithm that produced the components, which decides what their
      values and signs mean.
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
    required: true
    any_of:
    - range: DecompositionMethod
    - range: string
  second_block:
    name: second_block
    description: 'What the brain block was decomposed against, one value per additional
      block, and the primary filter on this payload. `[none]` for a decomposition
      of the brain data alone, which is a positive assertion rather than an empty
      slot.

      Multivalued because papers cross more than two blocks: a three-modality mCCA+jICA
      carries `other_imaging_modality` twice, and a design-plus-behaviour PLS carries
      `[design, behavioral]`. Name the blocks, never their number.'
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
    required: true
    multivalued: true
    any_of:
    - range: SecondBlock
    - range: string
  second_block_assessments:
    name: second_block_assessments
    description: The measures forming a behavioural or clinical block, by identifier,
      when they are `Assessment` records of this study. Empty when the second block
      is the design, another imaging modality, or anything else that is not an assessment
      -- those are named in `second_block_detail`.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: Assessment
    multivalued: true
    inlined: false
  second_block_detail:
    name: second_block_detail
    description: 'What each non-assessment block held, in the source''s words: the
      seed regions of a seed decomposition, the modalities of a fusion ("ALFF, FA,
      grey matter"), the genotype array of a parallel ICA ("367 SNPs"). Free text
      because the range of things that form a second block is open, and a SNP array
      is neither an `Assessment` nor an `Acquisition` of this study.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
    multivalued: true
  component_id:
    name: component_id
    description: Identifier of the component or latent variable within the decomposition,
      as reported -- commonly an index such as `IC12` or `LV1`. Required because without
      it two analyses from one decomposition are indistinguishable.
    in_subset:
    - mvp
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
    required: true
  component_label:
    name: component_label
    description: The source's name for the component, such as a network label, or
      for what a latent variable captures.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
  total_components:
    name: total_components
    description: Number of components or latent variables the decomposition produced.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: integer
    minimum_value: 0
  dimensionality_selection_method:
    name: dimensionality_selection_method
    description: How the number of components was chosen, as reported -- minimum description
      length, a variance-retained cutoff, a scree criterion, permutation on the singular
      values, or a fixed a priori number.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
  selection_criterion:
    name: selection_criterion
    description: How this component was selected for reporting.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
  variance_explained_percent:
    name: variance_explained_percent
    description: 'Percentage of the brain block''s variance this component accounts
      for, as reported. Not interchangeable with `crossblock_covariance_percent`:
      one is variance within a block and the other covariance between blocks, and
      averaging across the two would combine incomparable quantities.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: float
    minimum_value: 0
    maximum_value: 100
  crossblock_covariance_percent:
    name: crossblock_covariance_percent
    description: Percentage of crossblock covariance this latent variable accounts
      for, as reported. The nearest thing a two-block decomposition has to an effect
      size, and what a paper uses to justify reporting one latent variable and not
      another. Unset when `second_block` is `[none]`, there being no second block
      to cross.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: float
    minimum_value: 0
    maximum_value: 100
  map_quantity:
    name: map_quantity
    description: What the values of the reported map represent -- a loading, a weight,
      a PLS salience, a standardized score.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
    any_of:
    - range: MapQuantity
    - range: string
  salience_threshold:
    name: salience_threshold
    description: 'The criterion deciding which elements count as reliable contributors,
      as reported -- conventionally a bootstrap ratio cutoff such as "|BSR| > 3.0",
      or a standardized cutoff such as "|Z| > 3.5". Not `InferenceSettings`: a bootstrap
      ratio is a stability measure and not a test statistic, so reading it as a voxelwise
      threshold would misstate the inference.'
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: string
  polarity_semantics:
    name: polarity_semantics
    description: Whether the sign of the reported map carries meaning. A latent variable's
      sign is arbitrary up to a flip, and so is an independent component's, so a paper
      must say what positive means before a direction can be pooled.
    in_subset:
    - model_extracted
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    owner: LatentDecompositionDetails
    domain_of:
    - LatentDecompositionDetails
    range: PolaritySemantics
  details_type:
    name: details_type
    description: The concrete details subclass carried by this payload. Populated
      from the class itself, not extracted.
    from_schema: https://neurostuff.org/schemas/neuroimaging-study-storage/analysis_details
    rank: 1000
    designates_type: true
    owner: LatentDecompositionDetails
    domain_of:
    - AnalysisDetails
    range: string
    required: true