Data.yaml

# Important editor notes
# - Here [mappings](https://linkml.io/linkml/schemas/uris-and-mappings.html#mappings) indicate closely related templates in other data models
# - Here `extensions.suggests` are used to track the relationship between templates and their intended application to assay data types
# - `abstract` means the template is not meant to be used for annotation, it's just there for others to inherit props from

classes:

############################################################
# Assay Data / Sequencing             
############################################################

  BiologicalAssayDataTemplate:
    is_a: Template
    abstract: true
    description: >
      A template defining basic metadata on deposited data artifacts (i.e. files) from experimental assays involving biosamples. 
      This is an abstract template; "real" template subclasses define additional properties appropriate for the type of data file (e.g. imaging vs sequencing).
    slots:
    - Component
    - Filename
    - fileFormat
    - resourceType # should be constrained to data resource types but we can't do this yet
    - dataType
    - dataSubtype
    - assay # platform added at a later level, see https://github.com/nf-osi/nf-metadata-dictionary/issues/319
    - individualID # specimenID, aliquotID etc. are added at a later level since they don't make sense for some assay types such as behavioral assays
    - species # one day will these will be submitted separately
    - sex
    - age
    - ageUnit
    - diagnosis
    - nf1Genotype
    - nf2Genotype
    - tumorType
    - modelSystemName
    - organ
    - comments

  NonBiologicalAssayDataTemplate:
    is_a: Template
    abstract: true
    description: >
      Describes data artifacts (i.e. files) from an experimental/physical-sciences assay 
      where input specimens are more at the level of synthesized molecules or inorganic materials.
    notes:
    - Name of this base abstraction is not great, may need to revise.
    slots:
    - Component
    - Filename
    - fileFormat
    - resourceType
    - dataType
    - dataSubtype
    - assay
  
  #####################################################
  # Biological Assay Data / Sequencing / Raw            
  #####################################################

  GeneticsAssayTemplate:
    is_a: BiologicalAssayDataTemplate
    abstract: true
    description: > 
      Template for relatively raw data of RNA/DNA structure and expression. 
      This is an abstract template encapsulating data from low-throughput to high-throughput assays, 
      sequencing-based or non-sequencing based (e.g. microarrays, optical genome mapping). 
      In practice, data are more specifically typed and matched to one of the templates below.
    notes:
    - http://purl.obolibrary.org/obo/GECKO_0000032
    slots:
    - parentSpecimenID
    - specimenID 
    - aliquotID
    - tumorType 
    - platform
    - nucleicAcidSource
    - specimenPreparationMethod
    annotations:
      requiresComponent: ''
      required: false


  BulkSequencingAssayTemplate:
    is_a: GeneticsAssayTemplate
    description: General template for raw (level 1) RNA/DNA data, i.e. sequence data from a sequencing assay.
    slots:
    - specimenType
    - runType
    - libraryStrand
    - libraryPrep
    - libraryPreparationMethod
    - readPair
    - readLength
    - readDepth
    - targetDepth
    - batchID
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - ISO-seq
        - miRNA-seq
        - NOMe-seq
        - next-generation sequencing
        - ribo-seq
        - scCGI-seq
        - Sanger sequencing
        - T cell receptor repertoire sequencing

  ScSequencingAssayTemplate:
    is_a: GeneticsAssayTemplate
    description: General template for raw RNA/DNA data, i.e. sequence data from a sequencing assay.
    slots:
    - cellType
    - isCellLine
    - cellID
    - dissociationMethod
    - runType
    - libraryStrand
    - libraryKitID
    - libraryPrep
    - libraryPreparationMethod
    - readPair
    - readLength
    - readDepth
    - targetDepth
    - batchID
    - auxiliaryAsset
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - ISO-seq
        - miRNA-seq
        - NOMe-seq
        - next-generation sequencing
        - ribo-seq
        - scCGI-seq
        - Sanger sequencing
        - T cell receptor repertoire sequencing
        

  GenomicsAssayTemplate:
    is_a: BulkSequencingAssayTemplate
    description: Alias to BulkSequencingAssayTemplate, use for sequence data on a large scale when there is no template available that is more specific.
    notes:
    - It's possible that not all raw genomics data equates to sequencing output. 
    slots:
    annotations:
      requiresComponent: ''
      required: false


  GenomicsAssayTemplateExtended:
    is_a: GenomicsAssayTemplate
    description: Genomics assay template but with additional experiment data.
    slots:
    - experimentalCondition
    - experimentalTimepoint
    - timepointUnit
    - genePerturbed
    - genePerturbationTechnology
    - genePerturbationType
    - comments
    annotations:
      requiresComponent: ''
      required: false
  
  RNASeqTemplate:
    is_a: BulkSequencingAssayTemplate
    description: Template for describing raw data from (bulk) RNA-sequencing
    slots:
    - genePerturbed
    - genePerturbationType
    - genePerturbationTechnology
    - experimentalCondition
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - RNA-seq
        - lncRNA-seq
        dataType:
        - SequenceReads
    close_mappings:
    - htan:BulkRNA-seqLevel1

  ScRNASeqTemplate:
    is_a: ScSequencingAssayTemplate
    description: Template for describing raw data from single-cell RNA-seq.
    slots:
      - genePerturbed
      - genePerturbationType
      - genePerturbationTechnology
      - experimentalCondition
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - single-cell RNA-seq
        dataType:
        - geneExpression

    #Saving this for later/more time to think through
    #GeoMXSpatialTranscriptomicsTemplate:
    #is_a: GeneticsAssayTemplate
    #description: Files contain raw data output from the NanoString GeoMx DSP Pipeline. These can include RCC or DCC Files.
    #source: https://raw.githubusercontent.com/ncihtan/data-models/main/HTAN.model.csv
    #slots:
    #- libraryPrep
    #- libraryPreparationMethod
    #annotations:
    #  requiresComponent: ''
    #  required: false
    #  match:
    #    assay:
    #    - spatial transcriptomics
    #    dataType:
    #    - SequenceReads
    #close_mappings:
    #- htan:NanoStringGeoMxDSPSpatialTranscriptomicsLevel1
        
  PdxGenomicsAssayTemplate:
    is_a: GenomicsAssayTemplate
    description: Raw genomics data from patient-derived xenograft (PDX) experiment, with additional PDX-relevant metadata.
    slots:
    - transplantationType
    - transplantationRecipientSpecies
    - transplantationRecipientTissue
    - modelSystemName
    - experimentalCondition
    - experimentalTimepoint
    - timepointUnit
    annotations:
      requiresComponent: ''
      required: false
  
  WGSTemplate:
    is_a: BulkSequencingAssayTemplate
    description: Template for describing raw data from Whole Genome Sequencing (WGS)
    slots:
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - whole genome sequencing
        dataType:
        - SequenceReads
  
  WESTemplate:
    is_a: BulkSequencingAssayTemplate
    description: Template for describing raw data from Whole Exome Sequencing (WES/WXS)
    slots:
    - targetCaptureKitID
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - whole exome sequencing
        dataType:
        - SequenceReads
    close_mappings:
    - htan:BulkWESLevel1

  EpigenomiscAssayTemplate:
    is_a: BulkSequencingAssayTemplate
    description: Template for describing raw data from epigenetics sequencing assays such as bisulfite sequencing. 
    slots:
    - bisulfiteConversionKitID
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - ATAC-seq
        - bisulfite sequencing
        - CAPP-seq
        - CUT&RUN
        - ChIP-seq
        - ERR bisulfite sequencing
        - HI-C
        - NOMe-seq
        - oxBS-seq
    close_mappings:
    - htan:BulkMethylation-seqLevel1

  EpigeneticsAssayTemplate:
    is_a: EpigenomiscAssayTemplate
    description: Alias for EpigenomiscAssayTemplate for backwards-compatibility.
    slots:
    annotations:
      requiresComponent: ''
      required: false

  GenomicsArrayTemplate:
    is_a: GeneticsAssayTemplate
    description: A template for describing raw data from array-based genomics/epigenomics, e.g. CEL files.
    slots:
    - platform
    - channel
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay: 
        - array
        - methylation array
        - microarray
        - miRNA array
        - RNA array
        - SNP array
        - NanoString nCounter Analysis System

  MethylationArrayTemplate:
    is_a: GenomicsArrayTemplate
    description: Template for raw data files (idat) from DNA methylation arrays.
    source: https://docs.gdc.cancer.gov/Data_Dictionary/viewer/#?view=table-definition-view&id=raw_methylation_array
    slots:
    - chipID
    - chipPosition
    - plateName
    - plateWell
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay: 
        - methylation array


  #########################################################
  # Biological Assay Data  / Sequencing / Processed            
  #########################################################
  
  ProcessedAlignedReadsTemplate:
    is_a: BiologicalAssayDataTemplate
    description: >
      Template for describing aligned reads (e.g. BAM/CRAM files) from a sequencing assay.
      The QC meta are extracted from samtools stats when available and are the same metrics preferred by GDC. 
    slots:
    - genomicReference
    - genomicReferenceLink
    - averageInsertSize
    - averageReadLength
    - averageBaseQuality
    - pairsOnDifferentChr
    - readsDuplicatedPercent
    - readsMappedPercent
    - meanCoverage
    - proportionCoverage10x
    - proportionCoverage30x
    - readDepth
    - totalReads
    - workflow
    - workflowLink
    - auxiliaryAsset # more qc files
    - specimenID
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - ATAC-seq
        - CAPP-seq
        - CUT&RUN
        - ChIP-seq
        - ERR bisulfite sequencing
        - ISO-seq
        - NOMe-seq
        - RNA-seq
        - T cell receptor repertoire sequencing
        - bisulfite sequencing
        - miRNA-seq
        - next-generation sequencing
        - scCGI-seq
        - targeted exome sequencing
        - whole exome sequencing
        - whole genome sequencing
        dataType:
        - AlignedReads
    close_mappings:
    - htan:ScRNA-seqLevel2
    - htan:BulkRNA-seqLevel2
    - htan:BulkWESLevel2
  
  ProcessedVariantCallsTemplate:
    is_a: BiologicalAssayDataTemplate
    description: Template for describing either simple germline/somatic variant calls output data (VCF/MAF) as well as structural variants (e.g. CNVs).
    slots:
    - isFilteredReads
    - workflow
    - workflowLink
    - auxiliaryAsset
    - specimenID
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - next-generation sequencing
        - targeted exome sequencing
        - whole exome sequencing
        - whole genome sequencing
        - TIDE
        dataType:
        - GermlineVariants
        - AnnotatedGermlineVariants
        - SomaticVariants
        - AnnotatedSomaticVariants
        - CopyNumberVariants
    close_mappings:
    - htan:BulkWESLevel3

#################################################
# Biological Assay Data / Processed / Aggregated       
#################################################
    

  ProcessedExpressionTemplate:
    is_a: BiologicalAssayDataTemplate
    description: Template for quantified gene/protein expression data that are still represented as one file per sample.
    slots:
    - expressionUnit
    - workflow
    - workflowLink
    - auxiliaryAsset
    - specimenID
    annotations:
      requiresComponent: ''
      required: false  
      match:
        assay:
        - quantitative PCR
        - microarray
        - miRNA array
        - RNA array
        - RNA-seq
        - single-cell RNA-seq
        dataType:
        - geneExpression # .sf
        - proteinExpression
    close_mappings:
    - htan:BulkRNA-seqLevel3

  ProcessedMergedDataTemplate:
    description: > 
      Further processed data with multiple samples aggregated into one file.
      This may be also be known as level-4 data. Unlike level-2 and level-3 data, individual-level attributes such as age and sex are no longer surfaced on the data file directly.
    slots:
    - Component
    - Filename
    - fileFormat
    - resourceType
    - dataType
    - dataSubtype
    - workflow
    - workflowLink
    - auxiliaryAsset
    - assay
    annotations:
      requiresComponent: ''
      required: false  
      match:
        dataType:
        - AnnotatedSomaticVariants # merged MAFs
        - AnnotatedGermlineVariants # merged MAFs
        - geneExpresssion # TSVs
    close_mappings:
    - https://docs.gdc.cancer.gov/Data_Dictionary/viewer/#?view=table-definition-view&id=aggregated_somatic_mutation
    - https://docs.gdc.cancer.gov/Data_Dictionary/viewer/#?view=table-definition-view&id=gene_expression


##################################################
# Biological Assay Data / Proteomics     
##################################################

  ProteinAssayTemplate:
    is_a: BiologicalAssayDataTemplate
    abstract: true
    description: Abstract template for data from some assay of protein structure and function. Data should be instantiated with more specific template. 
    slots:
    - parentSpecimenID
    - specimenID
    - aliquotID
    - platform
    - proteinExtractSource
    annotations:
      requiresComponent: ''
      required: false


  MassSpecAssayTemplate:
    is_a: ProteinAssayTemplate
    description: Template for raw mass spec-based proteomics data.
    notes:
    - Technically, mass spec does not have to be considered a biological assay, but we'll classify it as such when used with a biological sample. 
    slots:
    - dataCollectionMode
    annotations:
      requiresComponent: ''
      required: false  
      match:
        assay:
        - mass spectrometry
        - FIA-MSMS
        - label free mass spectrometry
        - MudPIT
        - TMT quantitation
        - liquid chromatography/mass spectrometry
        - liquid chromatography/tandem mass spectrometry
        - ultra high-performance liquid chromatography/tandem mass spectrometry
    close_mappings:
    - htan:MassSpectrometryLevel1


  ProteomicsAssayTemplate:
    is_a: MassSpecAssayTemplate
    description: Alias to MassSpecAssayTemplate for backwards-compatibility.
    slots:
    annotations:
      requiresComponent: ''
      required: false  

  
  ProteinArrayTemplate:
    is_a: ProteinAssayTemplate
    description: Template for array- and immuno-based proteomics data.
    slots:
    - antibodyID
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - RPPA
    close_mappings:
    - htan:RPPALevel2


##############################################
# Biological Assay Data / Imaging / Raw           
##############################################

  ImagingAssayTemplate:
    is_a: BiologicalAssayDataTemplate
    description: General template for describing imaging data.
    slots:
    - platform
    - assayTarget
    - auxiliaryAsset
    annotations:
      requiresComponent: ''
      required: false
      match:
      - 3D imaging
      - high content screen # also under microscopy
      - high frequency ultrasound
      - in vivo bioluminescence
      - gel filtration chromatography
      - laser speckle imaging
      - photograph
      - positron emission tomography
      - spatial frequency domain imaging
      - western blot

  MRIAssayTemplate:
    is_a: ImagingAssayTemplate
    description: Template for describing MRI data.
    slots:
    - bodySite # should be constrained to brain regions
    - MRISequence
    - experimentalCondition
    - experimentalTimepoint
    - timepointUnit
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - conventional MRI
        - functional MRI
  
  MicroscopyAssayTemplate:
    is_a: ImagingAssayTemplate
    description: Template for data from a microscopy data.
    slots:
    - parentSpecimenID
    - specimenID 
    - aliquotID
    - objective
    - nominalMagnification
    - lensAperture
    - workingDistance
    - workingDistanceUnit
    - immersion
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - 3D confocal imaging
        - 3D electron microscopy
        - atomic force microscopy
        - brightfield microscopy
        - high content screen
        - phase-contrast microscopy
        - traction force microscopy
    close_mappings:
    - htan: ImagingLevel2

  ImmunoMicroscopyTemplate:
    is_a: MicroscopyAssayTemplate
    description: Template for describing immunofluorescence or immunohistochemistry images.
    slots:
    - antibodyID
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - immunofluorescence
        - immunohistochemistry  


############################################
# Biological Assay Data / Other           
############################################

  ClinicalAssayTemplate:
    is_a: BiologicalAssayDataTemplate
    description: > 
      General template for typically tabular **individual-level** data. This can include repeated measures and a drug treatment context.
    slots:
    - experimentalFactor
    - experimentalCondition
    - timepointUnit
    - compoundName
    - compoundDose
    - compoundDoseUnit
    - comments
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - blood chemistry measurement
        - body size trait measurement
        - survival
        - contextual conditioning behavior assay
        - Corsi blocks
        - elevated plus maze test
        - gait measurement
        - NIH Toolbox
        - questionnaire
        - Riccardi and Ablon scales
        - Social Responsiveness Scale
        - 'Social Responsiveness Scale, Second Edition'
        - Von Frey test
        - 'Child Behavior Checklist for Ages 1.5-5'
        - 'Child Behavior Checklist for Ages 6-18'
        dataType:
        - clinical

  GeneralMeasureDataTemplate:
    is_a: BiologicalAssayDataTemplate
    description: > 
      General template for data in tabular form that aggregates tissue-level or cellular-level data.
    slots:
    - parentSpecimenID
    - specimenID
    - aliquotID
    - modelSystemName
    - experimentalFactor
    - experimentalCondition
    - experimentalTimepoint
    - timepointUnit
    - compoundName
    - compoundDose
    - compoundDoseUnit
    - genePerturbed
    - genePerturbationType
    - genePerturbationTechnology
    - comments
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - cell competition
        - cell count
        - cell proliferation
        - cell viability assay
        - combination library screen
        - combination screen
        - complex II enzyme activity assay
        - compound screen
        - BrdU proliferation assay
        - electrochemiluminescence
        - focus forming assay
        - in vivo tumor growth
        - Matrigel-based tumorigenesis assay

  PharmacokineticsAssayTemplate:
    is_a: BiologicalAssayDataTemplate
    description: Generic template for describing data from a pharmacokinetics assay. 
    slots:
    - parentSpecimenID
    - specimenID
    - aliquotID
    - platform
    - compoundName
    - compoundDose
    - compoundDoseUnit
    - experimentalCondition
    - experimentalTimepoint
    - timepointUnit
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - 2D AlamarBlue absorbance
        - 2D AlamarBlue fluorescence
        - 2D Incucyte
        - ELISA
        - sandwich ELISA

  PlateBasedReporterAssayTemplate:
    is_a: BiologicalAssayDataTemplate
    description: Generic template for describing data from a plate-based reporter assay.
    slots:
    - parentSpecimenID
    - specimenID
    - aliquotID
    - platform
    - assayTarget
    - compoundName
    - compoundDose
    - compoundDoseUnit
    - experimentalCondition
    - experimentalTimepoint
    - timepointUnit
    - reporterGene
    - reporterSubstance
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - 2D AlamarBlue absorbance
        - 2D AlamarBlue fluorescence
        - 2D Incucyte
        - ELISA
        - sandwich ELISA

  FlowCytometryTemplate:
    is_a: BiologicalAssayDataTemplate
    description: Template for flow cytometry assay
    slots:
    - platform
    - cellType
    - auxiliaryAsset
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - flow cytometry
    close_mappings:
    - https://www.immport.org/shared/dataModelDocumentation?table=fcs_analyzed_result


  MaterialScienceAssayTemplate:
    is_a: NonBiologicalAssayDataTemplate
    description: General template for describing data for a materials science assay.
    slots:
    - platform
    - specimenID
    - materialType
    - concentrationMaterial
    - concentrationMaterialUnit
    - concentrationNaCl
    - concentrationNaClUnit
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - rheometry
        - oscillatory rheology
        - differential scanning calorimetry
  
  LightScatteringAssayTemplate:
    is_a: MaterialScienceAssayTemplate
    description: Template for dynamic or static light scattering data adapted from ISA-TAB-Nano specs.
    source: https://wiki.nci.nih.gov/display/icr/isa-tab-nano
    slots:
    - pH
    - concentrationNaCl
    - concentrationNaClUnit
    annotations:
      requiresComponent: ''
      required: false
      match:
        assay:
        - dynamic light scattering
        - microrheology
        - oscillatory rheology
        - static light scattering

  ElectrophysiologyAssayTemplate:
    is_a: BiologicalAssayDataTemplate
    description: Template for raw electrophysiology data (electrical recordings).
    source: https://arxiv.org/pdf/1605.07673.pdf
    slots:
    - platform
    - bodySite # applies for electrophysiology at the body level
    - cellType # applies for electrophysiology at the cell level, e.g. whole-cell patch clamp
    - recordingSource
    - experimentalTimepoint
    - timepointUnit
    - experimentalCondition
    - auxiliaryAsset
    match:
      assay:
      - local field potential recording
      - long term potentiation assay
      - multi-electrode array
      - pattern electroretinogram
      - whole-cell patch clamp


#######################################
# Other Data Artifacts           
#######################################

  WorkflowReport:
    is_a: Template
    description: Template used for miscellaneous workflow reports and accessory files
    notes:
    - resourceType should default to "workflow report"
    - note that `dataType` and `dataSubtype` don't apply
    slots:
    - resourceType
    - assay
    - fileFormat
    - relatedDataset
    - workflow
    - workflowLink


  UpdateMilestoneReport:
    is_a: PartialTemplate
    description: Metadata template for updating milestone report values in NF studies -- currently a supported feature for NTAP and GFF.
    slots:
    - Filename
    - resourceType
    - progressReportNumber
    annotations:
      requiresComponent: ''
      required: false