#! /usr/bin/env guile
!#

(use-modules (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 probeset
  (tables (ProbeSet
           (left-join GeneChip "ON GeneChip.Id = ProbeSet.ChipId")))
  (schema-triples
   (gnt:hasChip a owl:ObjectProperty)
   (gnt:hasChip rdfs:domain gnc:Probeset)
   (gnt:hasTargetId a owl:ObjectProperty)
   (gnt:hasTargetId rdfs:domain gnc:Probeset)
   (gnt:symbol rdfs:domain gnc:Probeset)
   (gnt:targetsRegion a owl:ObjectProperty)
   (gnt:targetsRegion rdfs:domain gnc:Probeset)
   (gnt:chr rdfs:domain gnc:Probeset)
   (gnt:mb rdfs:domain gnc:Probeset)
   (gnt:hasSpecificity a owl:ObjectProperty)
   (gnt:hasSpecificity rdfs:domain gnc:Probeset)
   (gnt:hasBlatScore a owl:ObjectProperty)
   (gnt:hasBlatScore rdfs:domain gnc:Probeset)
   (gnt:hasBlatMbStart a owl:ObjectProperty)
   (gnt:hasBlatMbStart rdfs:domain gnc:Probeset)
   (gnt:hasBlatMbEnd a owl:ObjectProperty)
   (gnt:hasBlatMbEnd rdfs:domain gnc:Probeset)
   (gnt:hasBlatSeq a owl:ObjectProperty)
   (gnt:hasBlatSeq rdfs:domain gnc:Probeset)
   (gnt:hasTargetSeq a owl:ObjectProperty)
   (gnt:hasTargetSeq rdfs:domain gnc:Probeset)
   (gnt:hasHomologeneId a owl:ObjectProperty)
   (gnt:hasHomologeneId rdfs:domain gnc:Probeset)
   (gnt:hasPubChemId a owl:ObjectProperty)
   (gnt:hasPubChemId rdfs:domain gnc:Probeset)
   (gnt:hasKeggId a owl:ObjectProperty)
   (gnt:hasKeggId rdfs:domain gnc:Probeset)
   (gnt:hasOmimId a owl:ObjectProperty)
   (gnt:hasOmimId rdfs:domain gnc:Probeset)
   (gnt:hasChebiId a owl:ObjectProperty)
   (gnt:hasChebiId rdfs:domain gnc:Probeset))
  (triples
      (let ((id (field ("IF(NULLIF(TRIM(ProbeSet.Name), '') IS NULL, '', TRIM(ProbeSet.Name))"
                        ProbeSetIdName)))
            (probeset-id (field ProbeSet Id)))
        (string->identifier
         "probeset"
         (if (string-null? id)
             (number->string probeset-id)
             (regexp-substitute/global
              #f "[^A-Za-z0-9:]"
              id
              'pre "_" 'post))))
    (set rdf:type 'gnc:Probeset)
    (set rdfs:label (field ProbeSet Name))
    (set skos:altLabel
         (replace-substrings
          (field ProbeSet alias)
          '(("\r\n" . "; "))))
    (set gnt:hasChip
         (string->identifier
          "platform"
          (field ("IFNULL(GeneChip.Name, '')" GeneChipName))))
    (set gnt:hasTargetId
         (field ("NULLIF(TRIM(ProbeSet.TargetId), '')"
                 TargetId)))
    (set gnt:symbol (field ProbeSet Symbol))
    (set dct:description (sanitize-rdf-string (field ProbeSet description)))
    (set gnt:targetsRegion
         (sanitize-rdf-string
          (field ("NULLIF(TRIM(ProbeSet.Probe_set_target_region), '')"
                  Probe_set_target_region))))
    (set gnt:chr (field ProbeSet Chr))
    (set gnt:mb (annotate-field (field ("IFNULL(ProbeSet.Mb, '')" Mb)) '^^xsd:double))
    (set gnt:hasSpecificity
         (field ("IFNULL(ProbeSet.Probe_set_specificity, '')"
                 Probe_set_specificity)))
    (set gnt:hasBlatScore
         (field ("IFNULL(ProbeSet.Probe_set_BLAT_score, '')"
                 Probe_set_BLAT_score)))
    (set gnt:hasBlatMbStart
         (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_start, '')"
                                 Probe_set_Blat_Mb_start))
                         '^^xsd:double))
    (set gnt:hasBlatMbEnd
         (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_end, '')"
                                 Probe_set_Blat_Mb_end))
                         '^^xsd:double))
    (set gnt:hasBlatSeq (sanitize-rdf-string (field ProbeSet BlatSeq)))
    (set gnt:hasTargetSeq (sanitize-rdf-string (field ProbeSet TargetSeq)))
    (set gnt:hasHomologeneId (ontology 'homologene:
                                       (field ("IFNULL(ProbeSet.HomoloGeneID, '')"
                                               HomoloGeneID))))
    (set gnt:hasUniprotId (ontology 'uniprot:
                                    (field ("IFNULL(ProbeSet.UniProtID, '')"
                                            UniProtID))))
    (set gnt:hasPubChemId (ontology
                           'pubchem:
                           (field ("IFNULL(ProbeSet.PubChem_ID, '')"
                                   PubChem_ID))))
    (set gnt:hasKeggId (ontology
                        'kegg:
                        (field ("IFNULL(ProbeSet.KEGG_ID, '')"
                                KEGG_ID))))
    (set gnt:hasOmimId (ontology
                        'omim:
                        (let ((omim (field ("IFNULL(ProbeSet.OMIM, '')"
                                            OMIM))))
                          (if (number? omim)
                              omim
                              (regexp-substitute/global
                               #f "[^0-9]"
                               omim
                               'pre "" 'post)))))
    (set gnt:hasChebiId (ontology
                         'chebi:
                         (field ("IFNULL(ProbeSet.ChEBI_ID, '')"
                                 ChEBI_ID))))))




(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 "ProbeSet Metadata")
   (connection %connection-settings)
   (table-metadata? #f)
   (prefixes
    '(("gn:" "<http://genenetwork.org/id/>")
      ("probeset:" "<http://genenetwork.org/probeset/>")
      ("gnc:" "<http://genenetwork.org/category/>")
      ("gnt:" "<http://genenetwork.org/term/>")
      ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>")
      ("kegg:" "<http://bio2rdf.org/ns/kegg#>")
      ("pubchem:" "<https://pubchem.ncbi.nlm.nih.gov/>")
      ("omim:" "<https://www.omim.org/entry/>")
      ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>")
      ("uniprot:" "<http://purl.uniprot.org/uniprot/>")
      ("chebi:" "<http://purl.obolibrary.org/obo/CHEBI_>")
      ("dct:" "<http://purl.org/dc/terms/>")
      ("owl:" "<http://www.w3.org/2002/07/owl#>")
      ("homologene:" "<https://bio2rdf.org/homologene:>")
      ("xsd:" "<http://www.w3.org/2001/XMLSchema#>")
      ("qb:" "<http://purl.org/linked-data/cube#>")
      ("sdmx-measure:" "<http://purl.org/linked-data/sdmx/2009/measure#>")
      ("skos:" "<http://www.w3.org/2004/02/skos/core#>")))
   (inputs
    (list probeset))
   (outputs
    `(#:documentation ,documentation
      #:rdf ,output))))