diff options
Diffstat (limited to 'examples/genelist.scm')
| -rwxr-xr-x | examples/genelist.scm | 200 |
1 files changed, 97 insertions, 103 deletions
diff --git a/examples/genelist.scm b/examples/genelist.scm index 9c1ced0..5048bf2 100755 --- a/examples/genelist.scm +++ b/examples/genelist.scm @@ -18,73 +18,72 @@ (tables (GeneList (left-join Species "USING (SpeciesId)"))) (schema-triples - (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:has_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 - #f "[^A-Za-z0-9:]" - (string-trim-both - (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)) + "gene" (normalize-string-field (string-trim-both + (field ("CONCAT_WS('_', GeneSymbol, GeneID, AlignID)" GENE_UID)))) + #:separator "_") + (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)) @@ -94,7 +93,7 @@ "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)) @@ -107,7 +106,7 @@ (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)) @@ -129,7 +128,7 @@ (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)) @@ -147,7 +146,7 @@ species "#goto=genereport&id=" geneId - "a gnc:biogpsLink")) + "a gnc:biogps_link")) ""))) (set dct:references (let ((geneId (field GeneList GeneID))) @@ -157,7 +156,7 @@ "<~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)) @@ -175,7 +174,7 @@ species (uri-encode (string-trim-both symbol)) - "a gnc:genemaniaLink")) + "a gnc:genemania_link")) ""))) (set dct:references (let ((symbol (field GeneList GeneSymbol))) @@ -186,7 +185,7 @@ "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))) @@ -197,7 +196,7 @@ "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))) @@ -208,7 +207,7 @@ "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))) @@ -219,33 +218,27 @@ "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 + (set gnt:tx_start (annotate-field (field GeneList TxStart) '^^xsd:double)) - (set gnt:TxEnd (annotate-field + (set gnt:tx_end (annotate-field (field GeneList TxEnd) '^^xsd:double)) - (set gnt:Strand (string-trim-both (field GeneList Strand))) + (set gnt:strand (string-trim-both (field GeneList Strand))) (set - gnt:belongsToSpecies - (string->identifier - "" - (remap-species-identifiers - (string-trim-both (field Species Name))) - #:separator "" - #:proc string-capitalize-first)) + gnt:has_species (string->identifier "" (remap-species-identifiers (field Species Fullname)))) (set 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 @@ -257,25 +250,26 @@ (if (number? gene-uid) (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))) + gene-uid) + #:separator "_")) + (set rdf:type 'gnc:gene) + (set gnt:has_species 'gn:Rattus_norvegicus) + (set gnt:gene_symbol (string-trim-both (field GeneList_rn33 geneSymbol))) (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))) @@ -295,7 +289,7 @@ "<~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)))) @@ -306,7 +300,7 @@ "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 (string-trim-both (field GeneList_rn33 geneSymbol)))) @@ -317,7 +311,7 @@ "https://www.gtexportal.org/home/gene/" (uri-encode (string-trim-both symbol)) - "a gnc:gtexLink")) + "a gnc:gtex_link")) ""))) (set dct:references (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol)))) @@ -328,7 +322,7 @@ "http://www.proteinatlas.org/search/" (uri-encode (string-trim-both symbol)) - "a gnc:proteinAtlasLink")) + "a gnc:protein_atlas_link")) ""))))) @@ -349,10 +343,10 @@ (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/>") + '(("gn:" "<http://rdf.genenetwork.org/v1/id/>") + ("probeset:" "<http://rdf.genenetwork.org/v1/probeset/>") + ("gnc:" "<http://rdf.genenetwork.org/v1/category/>") + ("gnt:" "<http://rdf.genenetwork.org/v1/term/>") ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>") ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>") ("dct:" "<http://purl.org/dc/terms/>") |