default_rules.ttl

@prefix rr:    <http://www.w3.org/ns/r2rml#> .
@prefix rml:   <http://semweb.mmlab.be/ns/rml#> .
@prefix ql:    <http://semweb.mmlab.be/ns/ql#> .
@prefix csvw:  <http://www.w3.org/ns/csvw#> .
@prefix xsd:   <http://www.w3.org/2001/XMLSchema#> .
@prefix vcfc:  <https://w3id.org/vcf-core/vocab#> .

# Default VCF-RDFizer RML rules, targeting the VCF Core vocabulary.
#
# Namespace
# ---------
# The target is https://w3id.org/vcf-core/vocab# (prefix vcfc:), which replaces
# the retired https://w3id.org/vcf-rdfizer/vocab# (prefix vcfr:). The retired
# namespace is not redirected: it serves a deprecation document in which every
# former term is linked to its successor. No version IRI is pinned; these rules
# target the namespace, whose release cadence is independent of the converter's.
#
# Expected TSV inputs produced by src/vcf_as_tsv.sh:
# - /data/tsv/<sample>.file_metadata.tsv
# - /data/tsv/<sample>.header_lines.tsv
# - /data/tsv/<sample>.records.tsv
# - /data/tsv/<sample>.sample_calls.tsv (header-only compatibility source)
# - /data/tsv/<sample>.sample_format_values.tsv (header-only compatibility source)
# The wrapper rewrites the template paths below for each input sample.
#
# What is here and what is not
# ----------------------------
# The division is a single rule: this mapping carries every field whose RDF
# datatype is the same for every row, and the wrapper's direct N-Triples
# emitters carry everything else. RML cannot choose an object datatype per row,
# and it cannot decompose one source cell into several resources without a
# materialized helper table whose size is the product of its dimensions.
#
# Emitted by the wrapper, not here:
# - ID, ALT, QUAL, FILTER and INFO. Each may be the VCF missing token, which the
#   vocabulary requires as "."^^vcfc:Null rather than a plain string, so the
#   datatype depends on the row.
# - fileDate. The published shape requires xsd:date, but ##fileDate has no
#   mandated format in the VCF specification, so the value must be inspected.
# - The structured header representation: header-line subclasses, the
#   vcfc:HeaderAttribute resources that vcfc:StructuredHeaderLineShape requires
#   on every ##INFO/##FORMAT/##FILTER/##ALT/##contig/##META/##SAMPLE/##PEDIGREE
#   line, and the INFO/FORMAT/FILTER/ALT/contig/META declaration properties.
#   These all live inside one angle-bracketed cell.
# - The allele layer: vcfc:ReferenceAllele and vcfc:AltAllele resources with
#   their index, kind, symbolic type and breakend components, parsed out of the
#   REF and ALT columns.
# - The structured INFO representation, its vcfc:FieldValueItem decomposition,
#   and the SV carriers derived from reserved INFO keys.
# - Genotype data. The wrapper selects exactly one direct sample emitter:
#   expanded SampleCall / FormatFieldValue / Genotype triples, or condensed
#   SampleSet / CohortCallMatrix / FormatValueVector triples. The helper tables
#   stay header-only in either mode.
#
# Ordinals
# --------
# vcfc:lineIndex and vcfc:recordIndex are typed xsd:integer, not
# xsd:positiveInteger. The SHACL profiles constrain every ordinal with
# sh:datatype xsd:integer plus a sh:minInclusive bound, and a SHACL datatype
# constraint compares the datatype IRI exactly. An xsd:integer literal
# satisfies both the shape and the ontology's xsd:positiveInteger range, since
# the value 1 belongs to that value space however it is written.
#
# VCF versions
# ------------
# VCF Core supplies one SHACL overlay per VCF version, each scoped by the file's
# vcfc:fileFormat. Most of this mapping is version-neutral, because the fixed
# columns and the header-line structure do not change between 4.1 and 4.5. The
# one version-dependent thing RML can carry is the optional vcfc:VCF4xFile
# subclass on the file subject, which is the sentinel in <#VCFFileMap> below.
#
# The version-dependent *semantics* -- which INFO keys flatten a tuple, whether
# that tuple repeats per ALT allele, and whether the local-allele and
# base-modification FORMAT families exist -- are decompositions of one source
# cell, so they live in the wrapper's emitters like every other decomposition.
#
# Output format:
# - The default conversion targets triples (N-Triples) without named graph terms.

<#VCFFileMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/file_metadata.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}" ;
    rr:class vcfc:VCFFile ;
    # Version overlay. The wrapper rewrites this sentinel per input to the
    # optional subclass for the version the file's ##fileformat line declares
    # -- vcfc:VCF41File through vcfc:VCF45File -- which activates that version's
    # SHACL gate. When the version has no conformance overlay (VCFv4.0, or a
    # missing or malformed line) the sentinel is rewritten to vcfc:VCFFile,
    # which is already asserted above, so the graph never claims a version gate
    # it cannot support and never gains an undefined term.
    rr:class vcfc:VCFVersionFile
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:hasHeader ;
    rr:objectMap [ rr:template "file://{SOURCE_FILE}#header" ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:fileFormat ;
    rr:objectMap [ rml:reference "FILE_FORMAT" ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:sourceSoftware ;
    rr:objectMap [ rml:reference "SOURCE_SOFTWARE" ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:referenceGenome ;
    rr:objectMap [ rml:reference "REFERENCE_GENOME" ]
  ] .

<#VCFHeaderMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/file_metadata.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#header" ;
    rr:class vcfc:VCFHeader
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:hasColumnHeader ;
    rr:objectMap [ rr:template "file://{SOURCE_FILE}#header/columns" ]
  ] .

# The mandatory #CHROM line. Its ordered sample columns are attached by the
# wrapper with vcfc:hasGenotypeColumns, because they live in the records TSV
# header rather than in a row of any source table.
<#ColumnHeaderLineMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/file_metadata.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#header/columns" ;
    rr:class vcfc:ColumnHeaderLine
  ] .

# NOTE:
# Use explicit link maps (template-based) instead of parentTriplesMap joins.
# In distributed RMLStreamer execution, joins can duplicate parent map triples.
<#VCFFileToRecordLinkMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/records.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}"
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:hasRecord ;
    rr:objectMap [
      rr:template "file://{SOURCE_FILE}#record/{ROW_ID}"
    ]
  ] .

<#VCFHeaderToHeaderLineLinkMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/header_lines.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#header"
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:hasHeaderLine ;
    rr:objectMap [
      rr:template "file://{SOURCE_FILE}#header/line/{HEADER_INDEX}"
    ]
  ] .

# Every '##' line as a base vcfc:HeaderLine. The wrapper adds the subclass, the
# ordered vcfc:HeaderAttribute resources and the declaration properties.
# vcfc:lineIndex is what the SPARQL SHACL profile uses to check that header
# indices are unique and that ##fileformat is line 1.
<#HeaderLineMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/header_lines.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#header/line/{HEADER_INDEX}" ;
    rr:class vcfc:HeaderLine
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:headerKey ;
    rr:objectMap [ rml:reference "HEADER_KEY" ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:headerValue ;
    rr:objectMap [ rml:reference "HEADER_VALUE" ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:lineIndex ;
    rr:objectMap [ rml:reference "HEADER_INDEX" ; rr:datatype xsd:integer ]
  ] .

# CHROM, POS and REF are the three fixed fields VCF 4.5 forbids from being the
# missing token, so their datatype is fixed for every row and they belong here.
# ID and ALT may be missing and are emitted by the wrapper.
<#VCFRecordMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/records.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#record/{ROW_ID}" ;
    rr:class vcfc:VCFRecord
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:chrom ;
    rr:objectMap [ rml:reference "CHROM" ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:pos ;
    rr:objectMap [ rml:reference "POS" ; rr:datatype xsd:integer ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:ref ;
    rr:objectMap [ rml:reference "REF" ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:recordIndex ;
    rr:objectMap [ rml:reference "ROW_ID" ; rr:datatype xsd:integer ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:hasCall ;
    rr:objectMap [ rr:template "file://{SOURCE_FILE}#call/{ROW_ID}" ]
  ] .

# FORMAT is the only call-level column whose datatype is fixed: VCF 4.5 defines
# it as a colon-separated key list with no missing form. QUAL, FILTER and INFO
# are emitted by the wrapper.
<#VariantCallMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/records.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#call/{ROW_ID}" ;
    rr:class vcfc:VariantCall
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:formatRaw ;
    rr:objectMap [ rml:reference "FORMAT" ]
  ] .

<#VariantCallToSampleLinkMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/sample_calls.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#call/{ROW_ID}"
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:hasSampleCall ;
    rr:objectMap [
      rr:template "file://{SOURCE_FILE}#sample/{ROW_ID}/{SAMPLE_URI_ID}"
    ]
  ] .

<#SampleCallMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/sample_calls.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#sample/{ROW_ID}/{SAMPLE_URI_ID}" ;
    rr:class vcfc:SampleCall
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:sampleId ;
    rr:objectMap [ rml:reference "SAMPLE_ID" ]
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:forSample ;
    rr:objectMap [
      rr:template "file://{SOURCE_FILE}#samples/{SAMPLE_URI_ID}"
    ]
  ] .

<#SampleCallToFormatValueLinkMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/sample_format_values.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#sample/{ROW_ID}/{SAMPLE_URI_ID}"
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:hasFormatValue ;
    rr:objectMap [
      rr:template "file://{SOURCE_FILE}#sample/{ROW_ID}/{SAMPLE_URI_ID}/fmt/{FORMAT_KEY}"
    ]
  ] .

<#FormatFieldValueMap>
  a rr:TriplesMap ;
  rml:logicalSource [
    rml:source [
      a csvw:Table ;
      csvw:url "/data/tsv/sample_format_values.tsv" ;
      csvw:dialect [
        a csvw:Dialect ;
        csvw:delimiter "\t" ;
        csvw:headerRowCount 1
      ]
    ] ;
    rml:referenceFormulation ql:CSV
  ] ;
  rr:subjectMap [
    rr:template "file://{SOURCE_FILE}#sample/{ROW_ID}/{SAMPLE_URI_ID}/fmt/{FORMAT_KEY}" ;
    rr:class vcfc:FormatFieldValue
  ] ;
  rr:predicateObjectMap [
    rr:predicate vcfc:fieldValue ;
    rr:objectMap [ rml:reference "FORMAT_VALUE" ]
  ] .