#! /usr/bin/env guile
!#

(use-modules (rnrs programs)
             (rnrs io ports)
             (srfi srfi-1)
             (srfi srfi-26)
             (ice-9 getopt-long)
             (ice-9 match)
             (ice-9 regex)
             (transform strings)
             (transform sql)
             (transform triples)
             (transform special-forms))


(define-transformer phenotypes
  (tables (PublishXRef
           (left-join InbredSet "ON InbredSet.InbredSetId = PublishXRef.InbredSetId")
           (left-join Publication "ON Publication.Id = PublishXRef.PublicationId")
           (left-join Phenotype "ON Phenotype.Id = PublishXRef.PhenotypeId")))
  (schema-triples
   (gnt:abbreviation a owl:ObjectProperty)
   (gnt:abbreviation rdfs:domain gnc:Phenotype)
   (gnt:abbreviation skos:definition "The abbreviation used for this resource")
   (gnt:labCode a owl:ObjectProperty)
   (gnt:labCode rdfs:domain gnc:Phenotype)
   (gnt:submitter a owl:ObjectProperty)
   (gnt:submitter rdfs:domain gnc:Phenotype)
   (gnt:submitter skos:definition "A person who submitted this resource to GN")
   (gnt:mean a rdf:Property)
   (gnt:mean a qb:MeasureProperty)
   (gnt:mean rdfs:subPropertyOf sdmx-measure:obsValue)
   (gnt:mean rdfs:domain gnc:Phenotype)
   (gnt:mean rdfs:range xsd:double)
   (gnt:lodScore a rdf:Property)
   (gnt:lodScore a qb:MeasureProperty)
   (gnt:lodScore rdfs:subPropertyOf sdmx-measure:obsValue)
   (gnt:lodScore rdfs:domain gnc:Phenotype)
   (gnt:lodScore rdfs:range xsd:double)
   (gnt:lodScore rdfs:label "Peak -logP")
   (gnt:lodScore skos:definition "Statistical measurement assessing the likelihood of genetic linkage between traits or genetic markers.")
   (gnt:locus a rdf:Property)
   (gnt:locus a qb:MeasureProperty)
   (gnt:locus rdfs:subPropertyOf sdmx-measure:obsValue)
   (gnt:locus rdfs:domain gnc:Phenotype)
   (gnt:locus rdfs:range rdfs:Literal)
   (gnt:additive rdfs:domain gnc:Phenotype)
   (gnt:additive rdfs:range xsd:double)
   (gnt:sequence rdfs:domain gnc:Phenotype)
   (gnt:sequence rdfs:range xsd:integer))
  (triples (string->identifier
            "trait"
            (field ("CONCAT(IFNULL(InbredSet.InbredSetCode, PublishXRef.InbredSetId), '_', PublishXRef.Id)"
                    Phenotype)))
    (set rdf:type 'gnc:Phenotype)
    (set gnt:belongsToGroup
         (string->identifier
          "set" (field InbredSet Name InbredSetName)
          #:separator ""
          #:proc string-capitalize-first))
    ;; This is the trait's name
    (set rdfs:label
         (let ((trait-id (field PublishXRef Id)))
           (if (number? trait-id)
               (number->string trait-id)
               trait-id)))
    (set skos:altLabel
         (field ("CONCAT(IFNULL(InbredSet.InbredSetCode, PublishXRef.InbredSetId), '_', PublishXRef.Id)"
                 Phenotype)))
    ;; All phenotypes have a post-publication description
    (set dct:description
         (sanitize-rdf-string
          (field Phenotype Post_publication_description)))
    ;; All phenotypes have a post-publication abbreviation
    (set gnt:abbreviation (field Phenotype Post_publication_abbreviation))
    (set gnt:labCode (field Phenotype Lab_code))
    (set gnt:submitter
         (sanitize-rdf-string (field Phenotype Submitter)))
    (multiset dct:contributor
              (map string-trim-both
                   (string-split
                    (sanitize-rdf-string (field Phenotype Owner))
                    #\,)))
    (set gnt:mean (annotate-field (field ("IFNULL(PublishXRef.mean, '')" mean))
                                  '^^xsd:double))
    (set gnt:locus
         (string->identifier
          ""
          (regexp-substitute/global
           #f "[^A-Za-z0-9:]"
           (sanitize-rdf-string (field PublishXRef Locus))
           'pre "_" 'post)
          #:separator ""
          #:proc string-capitalize-first))
    (set gnt:lodScore (annotate-field
                  (field ("IFNULL((PublishXRef.LRS/4.604), '')" lrs))
                  '^^xsd:double))
    (set gnt:additive
         (annotate-field (field ("IFNULL(PublishXRef.additive, '')" additive))
                         '^^xsd:double))
    (set gnt:sequence (annotate-field (field PublishXRef Sequence) '^^xsd:integer))
    (set dct:isReferencedBy
         (let ((pmid (field
                      ("IF(Publication.PubMed_ID IS NULL, '', CONVERT(Publication.PubMed_Id, INT))"
                       pmid)))
               (publication-id (field Publication Id PublicationId)))
           (if (string-null? pmid)
               (string->identifier "unpublished"
                                   (number->string publication-id))
               (ontology 'pubmed: pmid))))))



(let* ((option-spec
        '((settings (single-char #\s) (value #t))
          (output (single-char #\o) (value #t))
          (documentation (single-char #\d) (value #t))))
       (options (getopt-long (command-line) option-spec))
       (settings (option-ref options 'settings #f))
       (output (option-ref options 'output #f))
       (documentation (option-ref options 'documentation #f))
       (%connection-settings
        (call-with-input-file settings
          read)))
  (with-documentation
   (name "Phenotypes Metadata")
   (connection %connection-settings)
   (table-metadata? #f)
   (prefixes
    '(("dct:" "<http://purl.org/dc/terms/>")
      ("gn:" "<http://genenetwork.org/id/>")
      ("owl:" "<http://www.w3.org/2002/07/owl#>")
      ("gnc:" "<http://genenetwork.org/category/>")
      ("gnt:" "<http://genenetwork.org/term/>")
      ("sdmx-measure:" "<http://purl.org/linked-data/sdmx/2009/measure#>")
      ("skos:" "<http://www.w3.org/2004/02/skos/core#>")
      ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>")
      ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>")
      ("xsd:" "<http://www.w3.org/2001/XMLSchema#>")
      ("qb:" "<http://purl.org/linked-data/cube#>")
      ("xkos:" "<http://rdf-vocabulary.ddialliance.org/xkos#>")
      ("pubmed:" "<http://rdf.ncbi.nlm.nih.gov/pubmed/>")))
   (inputs
    (list
     phenotypes))
   (outputs
    `(#:documentation ,documentation
      #:rdf ,output))))