Update probeset dump

Signed-off-by: Munyoki Kilyungi <me@bonfacemunyoki.com>

1 files changed, 149 insertions, 22 deletions

diff --git a/examples/dump-probeset.scm b/examples/dump-probeset.scm
index 4d5f9a5..a9b6c7c 100755
--- a/examples/dump-probeset.scm
+++ b/examples/dump-probeset.scm
@@ -21,29 +21,144 @@
   (tables (ProbeSet
            (left-join GeneChip "ON GeneChip.Id = ProbeSet.ChipId")))
   (schema-triples
-   (gnt:name rdfs:range rdfs:Literal)
-   (gnt:probeset rdfs:range rdfs:Literal))
-  (triples (ontology
-            'probeset:
-            (string-trim-both
+   (gnc:probeset a skos:Concept)
+   (gnc:probeset
+    skos:description
+    "This is a set of controlled terms that are used to describe a given probeset")
+   (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:mbMm8 rdfs:domain gnc:probeset)
+   (gnt:mb2016 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:hasBlatMbStart2016 a owl:ObjectProperty)
+   (gnt:hasBlatMbStart2016 rdfs:domain gnc:probeset)
+   (gnt:hasBlatMbEnd a owl:ObjectProperty)
+   (gnt:hasBlatMbEnd rdfs:domain gnc:probeset)
+   (gnt:hasBlatMbEnd2016 a owl:ObjectProperty)
+   (gnt:hasBlatMbEnd2016 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)))
+        (if (string-null? id)
+            (string->identifier
+             "probeset"
+             (number->string
+              probeset-id))
+            (string->identifier
+             ""
              (regexp-substitute/global
               #f "[^A-Za-z0-9:]"
-              (field ("IFNULL(NULLIF(TRIM(ProbeSet.Name), ''), ProbeSet.Id)"
-                      name))
-              'pre "_" 'post)))
-    (set rdf:type 'gn-id:probeset)
-    (set gnt:chipOf (string->identifier "platform" (field GeneChip Name)))
-    (set gnt:name (field ProbeSet Name))
-    (set gnt:symbol (delete-substrings (field ProbeSet Symbol) "\""))
-    (set gnt:description (sanitize-rdf-string
-                              (field ProbeSet description)))
+              id
+              'pre "_" 'post)
+             #:separator ""
+             #:proc string-capitalize-first)))
+    (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:blatSeq (sanitize-rdf-string
-                          (string-trim-both (field ProbeSet BlatSeq))))
-    (set gnt:targetSeq (sanitize-rdf-string (field ProbeSet TargetSeq)))
-    (set gnt:uniProtReference (ontology 'uniprot:
-                                            (field ProbeSet UniProtID)))))
+    (set gnt:mbMm8 (annotate-field (field ("IFNULL(ProbeSet.Mb_mm8, '')" Mb_mm8))
+                                   '^^xsd:double))
+    (set gnt:mb2016
+         (annotate-field (field ("IFNULL(ProbeSet.Mb_2016, '')" Mb_2016))
+                         '^^xsd:double))
+    (set gnt:hasSpecificity
+         (annotate-field (field ("IFNULL(ProbeSet.Probe_set_specificity, '')"
+                                 Probe_set_specificity))
+                         '^^xsd:double))
+    (set gnt:hasBlatScore
+         (annotate-field (field ("IFNULL(ProbeSet.Probe_set_BLAT_score, '')"
+                                 Probe_set_BLAT_score))
+                         '^^xsd:double))
+    (set gnt:hasBlatMbStart
+         (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_start, '')"
+                                 Probe_set_Blat_Mb_start))
+                         '^^xsd:double))
+    (set gnt:hasBlatMbStart2016
+         (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_start_2016, '')"
+                                 Probe_set_Blat_Mb_start_2016))
+                         '^^xsd:double))
+    (set gnt:hasBlatMbEnd
+         (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_end, '')"
+                                 Probe_set_Blat_Mb_end))
+                         '^^xsd:double))
+    (set gnt:hasBlatMbEnd2016
+         (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_start_2016, '')"
+                                 Probe_set_Blat_Mb_start_2016))
+                         '^^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))))))
 
 
 
@@ -53,10 +168,22 @@
  (connection %connection-settings)
  (table-metadata? #f)
  (prefixes
-  '(("probeset:" "<http://genenetwork.org/probeset/>")
+  '(("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#>")
-    ("rdfs:" "<http://www.w3.org/2000/01/rdf-r")
-    ("uniprot:" "<http://purl.uniprot.org/uniprot/>")))
+    ("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#>")
+    ("skos:" "<http://www.w3.org/2004/02/skos/core#>")))
  (inputs
   (list dump-probeset))
  (outputs