about summary refs log tree commit diff
path: root/examples/genelist.scm
diff options
context:
space:
mode:
authorMunyoki Kilyungi2025-12-23 12:06:06 +0300
committerMunyoki Kilyungi2026-01-13 12:02:49 +0300
commit1ca7e679b834ccaf53a3243d0e1c2f3f9e8d56d8 (patch)
tree514c544706986f3edd0b3f53a89113e334a0b9a3 /examples/genelist.scm
parentc42933e8f474d8d14eac387d5a94da6f52210629 (diff)
downloadgn-transform-databases-1ca7e679b834ccaf53a3243d0e1c2f3f9e8d56d8.tar.gz
Snake case gn/gnt/gnc identifiers.
Signed-off-by: Munyoki Kilyungi <me@bonfacemunyoki.com>
Diffstat (limited to 'examples/genelist.scm')
-rwxr-xr-xexamples/genelist.scm190
1 files changed, 95 insertions, 95 deletions
diff --git a/examples/genelist.scm b/examples/genelist.scm
index 8729626..18fd30b 100755
--- a/examples/genelist.scm
+++ b/examples/genelist.scm
@@ -18,63 +18,63 @@
   (tables (GeneList
            (left-join Species "USING (SpeciesId)")))
   (schema-triples
-   (gnc:GeneSymbol a rdfs:Class)
-   (gnc:GeneSymbol rdfs:label "A gene symbol")
-   (gnt:gene rdfs:domain gnc:GeneSymbol)
-   (gnt:belongsToSpecies rdfs:domain gnc:GeneSymbol)
-   (gnc:Gene a rdfs:Class)
-   (gnc:Gene rdfs:label "Gene")
-   (gnt:hasGeneId a owl:ObjectProperty)
-   (gnt:hasGeneId rdfs:domain gnc:NCBIWikiEntry)
-   (gnt:hasGeneId skos:definition "The GeneId of this this resource")
-   (gnc:transcript rdfs:domain gnc:GeneSymbol)
+   (gnc:gene_symbol a rdfs:Class)
+   (gnc:gene_symbol rdfs:label "A gene symbol")
+   (gnt:gene rdfs:domain gnc:gene_symbol)
+   (gnt:belongs_to_species rdfs:domain gnc:gene_symbol)
+   (gnc:gene a rdfs:Class)
+   (gnc:gene rdfs:label "Gene")
+   (gnt:has_gene_id a owl:ObjectProperty)
+   (gnt:has_gene_id rdfs:domain gnc:ncbi_wiki_entry)
+   (gnt:has_gene_id skos:definition "The GeneId of this this resource")
+   (gnc:transcript rdfs:domain gnc:gene_symbol)
    (gnt:transcript a owl:ObjectProperty)
    (gnc:transcript rdfs:comments "The gene transcript of this resource")
-   (gnc:ebiGwasLink rdfs:Class gnc:ResourceLink)
-   (gnc:ebiGwasLink rdfs:label "EBI GWAS")
-   (gnc:ebiGwasLink rdfs:comments "EBI GWAS")
-   (gnc:proteinAtlasLink rdfs:Class gnc:ResourceLink)
-   (gnc:proteinAtlasLink rdfs:label "Protein Atlas")
-   (gnc:proteinAtlasLink rdfs:comments "Human Protein Atlas")
-   (gnc:genemaniaLink rdfs:Class gnc:ResourceLink)
-   (gnc:genemaniaLink rdfs:label "GeneMANIA")
-   (gnc:genemaniaLink rdfs:comments "GeneMANIA")
-   (gnc:gemmaLink rdfs:Class gnc:ResourceLink)
-   (gnc:gemmaLink rdfs:label "Gemma")
-   (gnc:gemmaLink rdfs:comments "Meta-analysis of gene expression data")
-   (gnc:biogpsLink rdfs:Class gnc:ResourceLink)
-   (gnc:biogpsLink rdfs:label "BioGPS")
-   (gnc:biogpsLink rdfs:comments "Expression across many tissues and cell types")
-   (gnc:abaLink rdfs:Class gnc:ResourceLink)
-   (gnc:abaLink rdfs:label "ABA")
-   (gnc:abaLink rdfs:comments "Allen Brain Atlas")
-   (gnc:pantherLink rdfs:Class gnc:ResourceLink)
-   (gnc:pantherLink rdfs:label "PANTHER")
-   (gnc:pantherLink rdfs:comments "Gene and protein data resources from Celera-ABI")
-   (gnc:stringLink rdfs:Class gnc:ResourceLink)
-   (gnc:stringLink rdfs:label "STRING")
-   (gnc:stringLink rdfs:comments "Protein interactions: known and inferred")
-   (gnc:gtexLink rdfs:Class gnc:ResourceLink)
-   (gnc:gtexLink rdfs:label "GTEx Portal")
-   (gnc:gtexLink rdfs:comments "GTEx Portal")
-   (gnc:rgdLink rdfs:Class gnc:ResourceLink)
-   (gnc:rgdLink rdfs:label "Rat Genome DB")
-   (gnc:rgdLink rdfs:comments "Rat Genome DB")
-   (gnc:hasKgID rdfs:domain gnc:GeneSymbol)
-   (gnt:hasKgID a owl:ObjectProperty)
-   (gnc:hasKgID rdfs:comments "The kgID of this resource")
-   (gnc:hasUnigenID rdfs:domain gnc:GeneSymbol)
-   (gnt:hasUnigenID a owl:ObjectProperty)
-   (gnc:hasUnigenID rdfs:comments "The UnigenID of this resource")
-   (gnc:hasProteinID rdfs:domain gnc:GeneSymbol)
-   (gnt:hasProteinID a owl:ObjectProperty)
-   (gnc:hasProteinID rdfs:comments "The ProteinID of this resource")
-   (gnc:hasAlignID rdfs:domain gnc:GeneSymbol)
-   (gnt:hasAlignID a owl:ObjectProperty)
-   (gnc:hasAlignID rdfs:comments "The AlignID of this resource")
-   (gnt:TxEnd rdfs:range xsd:double)
-   (gnt:TxStart rdfs:range xsd:double)
-   (gnt:hasTargetSeq rdfs:domain gnc:Probeset))
+   (gnc:ebi_gwas_link rdfs:Class gnc:ResourceLink)
+   (gnc:ebi_gwas_link rdfs:label "EBI GWAS")
+   (gnc:ebi_gwas_link rdfs:comments "EBI GWAS")
+   (gnc:protein_atlas_link rdfs:Class gnc:ResourceLink)
+   (gnc:protein_atlas_link rdfs:label "Protein Atlas")
+   (gnc:protein_atlas_link rdfs:comments "Human Protein Atlas")
+   (gnc:genemania_link rdfs:Class gnc:ResourceLink)
+   (gnc:genemania_link rdfs:label "GeneMANIA")
+   (gnc:genemania_link rdfs:comments "GeneMANIA")
+   (gnc:gemma_link rdfs:Class gnc:ResourceLink)
+   (gnc:gemma_link rdfs:label "Gemma")
+   (gnc:gemma_link rdfs:comments "Meta-analysis of gene expression data")
+   (gnc:biogps_link rdfs:Class gnc:ResourceLink)
+   (gnc:biogps_link rdfs:label "BioGPS")
+   (gnc:biogps_link rdfs:comments "Expression across many tissues and cell types")
+   (gnc:aba_link rdfs:Class gnc:ResourceLink)
+   (gnc:aba_link rdfs:label "ABA")
+   (gnc:aba_link rdfs:comments "Allen Brain Atlas")
+   (gnc:panther_link rdfs:Class gnc:ResourceLink)
+   (gnc:panther_link rdfs:label "PANTHER")
+   (gnc:panther_link rdfs:comments "Gene and protein data resources from Celera-ABI")
+   (gnc:panther_link rdfs:Class gnc:ResourceLink)
+   (gnc:panther_link rdfs:label "STRING")
+   (gnc:panther_link rdfs:comments "Protein interactions: known and inferred")
+   (gnc:gtex_link rdfs:Class gnc:ResourceLink)
+   (gnc:gtex_link rdfs:label "GTEx Portal")
+   (gnc:gtex_link rdfs:comments "GTEx Portal")
+   (gnc:rgd_link rdfs:Class gnc:ResourceLink)
+   (gnc:rgd_link rdfs:label "Rat Genome DB")
+   (gnc:rgd_link rdfs:comments "Rat Genome DB")
+   (gnc:has_kg_id rdfs:domain gnc:gene_symbol)
+   (gnc:has_kg_id a owl:ObjectProperty)
+   (gnc:has_kg_id rdfs:comments "The kgID of this resource")
+   (gnc:has_unigen_id rdfs:domain gnc:gene_symbol)
+   (gnc:has_unigen_id a owl:ObjectProperty)
+   (gnc:has_unigen_id rdfs:comments "The UnigenID of this resource")
+   (gnc:has_protein_id rdfs:domain gnc:gene_symbol)
+   (gnt:has_protein_id a owl:ObjectProperty)
+   (gnc:has_protein_id rdfs:comments "The ProteinID of this resource")
+   (gnc:has_align_id rdfs:domain gnc:gene_symbol)
+   (gnt:has_align_id a owl:ObjectProperty)
+   (gnc:has_align_id rdfs:comments "The AlignID of this resource")
+   (gnt:tx_end rdfs:range xsd:double)
+   (gnt:tx_start rdfs:range xsd:double)
+   (gnt:has_target_seq rdfs:domain gnc:probeset))
   (triples
       (string->identifier
        "gene" (regexp-substitute/global
@@ -83,10 +83,10 @@
                 (field ("CONCAT_WS('_', GeneSymbol, GeneID, AlignID)" GENE_UID)))
                'pre "_" 'post)
        #:proc (lambda (x) x))
-    (set rdf:type 'gnc:Gene)
-    (set gnt:geneSymbol (field GeneList GeneSymbol))
+    (set rdf:type 'gnc:gene)
+    (set gnt:gene_symbol (field GeneList GeneSymbol))
     (set dct:description (sanitize-rdf-string (field GeneList GeneDescription)))
-    (set gnt:hasGeneId (ontology 'gene: (field GeneList GeneId)))
+    (set gnt:has_gene_id (ontology 'gene: (field GeneList GeneId)))
     (set dct:references
          (let ((symbol (field GeneList GeneSymbol)))
            (if (not (string-blank? symbol))
@@ -96,7 +96,7 @@
                         "https://www.ebi.ac.uk/gwas/search?query="
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:ebiGwasLink"))
+                        "a gnc:ebi_gwas_link"))
                "")))
     (set dct:references
          (let ((symbol (field GeneList GeneSymbol))
@@ -109,7 +109,7 @@
                (string->symbol
                 (format #f "<~0@*~a> .~%<~0@*~a> ~1@*~a"
                         "http://mouse.brain-map.org/search/show?search_type=gene&search_term="
-                        "a gnc:abaLink"
+                        "a gnc:aba_link"
                         (if (string=? species "mouse")
                             (uri-encode
                              (string-trim-both symbol))
@@ -131,7 +131,7 @@
                          (string-trim-both symbol))
                         "&category=Gene&species="
                         (string-capitalize species)
-                        "a gnc:rgdLink"))
+                        "a gnc:rgd_link"))
                "")))
     (set dct:references
          (let ((geneId (field GeneList GeneID))
@@ -149,7 +149,7 @@
                         species
                         "#goto=genereport&id="
                         geneId
-                        "a gnc:biogpsLink"))
+                        "a gnc:biogps_link"))
                "")))
     (set dct:references
          (let ((geneId (field GeneList GeneID)))
@@ -159,7 +159,7 @@
                         "<~0@*~a~1@*~a> .~%<~0@*~a~1@*~a> ~2@*~a"
                         "http://www.chibi.ubc.ca/Gemma/gene/showGene.html?ncbiid="
                         geneId
-                        "a gnc:gemmaLink"))
+                        "a gnc:gemma_link"))
                "")))
     (set dct:references
          (let ((symbol (field GeneList GeneSymbol))
@@ -177,7 +177,7 @@
                         species
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:genemaniaLink"))
+                        "a gnc:genemania_link"))
                "")))
     (set dct:references
          (let ((symbol (field GeneList GeneSymbol)))
@@ -188,7 +188,7 @@
                         "http://www.pantherdb.org/genes/geneList.do?searchType=basic&fieldName=all&organism=all&listType=1&fieldValue="
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:pantherLink"))
+                        "a gnc:panther_link"))
                "")))
     (set dct:references
          (let ((symbol (field GeneList GeneSymbol)))
@@ -199,7 +199,7 @@
                         "http://string-db.org/newstring_cgi/show_network_section.pl?identifier="
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:stringLink"))
+                        "a gnc:panther_link"))
                "")))
     (set dct:references
          (let ((symbol (field GeneList GeneSymbol)))
@@ -210,7 +210,7 @@
                         "https://www.gtexportal.org/home/gene/"
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:gtexLink"))
+                        "a gnc:gtex_link"))
                "")))
     (set dct:references
          (let ((symbol (field GeneList GeneSymbol)))
@@ -221,18 +221,18 @@
                         "http://www.proteinatlas.org/search/"
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:proteinAtlasLink"))
+                        "a gnc:protein_atlas_link"))
                "")))
     (set gnt:chromosome (field GeneList Chromosome))
-    (set gnt:TxStart (annotate-field
-                      (field GeneList TxStart)
+    (set gnt:tx_start (annotate-field
+                      (field GeneList tx_start)
                       '^^xsd:double))
-    (set gnt:TxEnd (annotate-field
-                    (field GeneList TxEnd)
+    (set gnt:tx_end (annotate-field
+                    (field GeneList tx_end)
                     '^^xsd:double))
-    (set gnt:Strand (string-trim-both (field GeneList Strand)))
+    (set gnt:strand (string-trim-both (field GeneList Strand)))
     (set
-     gnt:belongsToSpecies
+     gnt:belongs_to_species
      (string->identifier
       ""
       (remap-species-identifiers
@@ -243,11 +243,11 @@
      gnt:transcript
      (ontology 'transcript:
                (string-trim-both (field GeneList NM_ID))))
-    (set gnt:hasKgID (string-trim-both (field GeneList kgID)))
-    (set gnt:hasUnigenID (string-trim-both (field GeneList UnigenID)))
-    (set gnt:hasProteinID (string-trim-both (field GeneList ProteinID)))
-    (set gnt:hasAlignID (string-trim-both (field GeneList AlignID)))
-    (set gnt:hasRgdID
+    (set gnc:has_kg_id (string-trim-both (field GeneList kgID)))
+    (set gnc:has_unigen_id (string-trim-both (field GeneList UnigenID)))
+    (set gnt:has_protein_id (string-trim-both (field GeneList ProteinID)))
+    (set gnt:has_align_id (string-trim-both (field GeneList AlignID)))
+    (set gnt:has_rgd_id
          (field ("IFNULL(RGD_ID, '')" RGD_ID)))))
 
 (define-transformer genelist-rn33
@@ -260,27 +260,27 @@
              (number->string
               gene-uid)
              gene-uid)))
-    (set rdf:type 'gnc:Gene)
-    (set gnt:belongsToSpecies 'gn:Rattus_norvegicus)
-    (set gnt:geneSymbol (string-trim-both (field GeneList_rn33 geneSymbol)))
+    (set rdf:type 'gnc:gene)
+    (set gnt:belongs_to_species 'gn:Rattus_norvegicus)
+    (set gnt:gene_symbol (string-trim-both (field GeneList_rn33 gene_symbol)))
     (set gnt:chromosome (field GeneList_rn33 chromosome))
-    (set gnt:TxStart (annotate-field
+    (set gnt:tx_start (annotate-field
                       (field GeneList_rn33 txStart)
                       '^^xsd:double))
-    (set gnt:TxEnd (annotate-field
+    (set gnt:tx_end (annotate-field
                     (field GeneList_rn33 txEnd)
                     '^^xsd:double))
-    (set gnt:Strand (string-trim-both (field GeneList_rn33 strand)))
+    (set gnt:strand (string-trim-both (field GeneList_rn33 strand)))
     (set
      gnt:transcript
      (ontology
       'transcript:
       (string-trim-both (field GeneList_rn33 NM_ID))))
     (set
-     gnt:hasKgID
+     gnc:has_kg_id
      (string-trim-both (field GeneList_rn33 kgID)))
     (set dct:references
-         (let ((symbol (field GeneList_rn33 geneSymbol)))
+         (let ((symbol (field GeneList_rn33 gene_symbol)))
            (if (not (string-blank? symbol))
                (string->symbol
                 (format #f
@@ -290,17 +290,17 @@
                         "a gnc:PantherLink"))
                "")))
     (set dct:references
-         (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol))))
+         (let ((symbol (string-trim-both (field GeneList_rn33 gene_symbol))))
            (if (not (string-blank? symbol))
                (string->symbol
                 (format #f
                         "<~0@*~a~1@*~a> .~%<~0@*~a~1@*~a> ~2@*~a"
                         "https://www.ebi.ac.uk/gwas/search?query="
                         (string-trim-both symbol)
-                        "a gnc:ebiGwasLink"))
+                        "a gnc:ebi_gwas_link"))
                "")))
     (set dct:references
-         (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol))))
+         (let ((symbol (string-trim-both (field GeneList_rn33 gene_symbol))))
            (if (not (string-blank? symbol))
                (string->symbol
                 (format #f
@@ -308,10 +308,10 @@
                         "http://string-db.org/newstring_cgi/show_network_section.pl?identifier="
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:stringLink"))
+                        "a gnc:panther_link"))
                "")))
     (set dct:references
-         (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol))))
+         (let ((symbol (string-trim-both (field GeneList_rn33 gene_symbol))))
            (if (not (string-blank? symbol))
                (string->symbol
                 (format #f
@@ -319,10 +319,10 @@
                         "https://www.gtexportal.org/home/gene/"
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:gtexLink"))
+                        "a gnc:gtex_link"))
                "")))
     (set dct:references
-         (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol))))
+         (let ((symbol (string-trim-both (field GeneList_rn33 gene_symbol))))
            (if (not (string-blank? symbol))
                (string->symbol
                 (format #f
@@ -330,7 +330,7 @@
                         "http://www.proteinatlas.org/search/"
                         (uri-encode
                          (string-trim-both symbol))
-                        "a gnc:proteinAtlasLink"))
+                        "a gnc:protein_atlas_link"))
                "")))))