diff options
| -rwxr-xr-x | examples/classification.scm | 27 | ||||
| -rwxr-xr-x | examples/dataset-metadata.scm | 113 | ||||
| -rwxr-xr-x | examples/genbank.scm | 4 | ||||
| -rwxr-xr-x | examples/genelist.scm | 36 | ||||
| -rwxr-xr-x | examples/generif.scm | 50 | ||||
| -rwxr-xr-x | examples/genotype.scm | 9 | ||||
| -rwxr-xr-x | examples/phenotype.scm | 16 | ||||
| -rwxr-xr-x | examples/schema.scm | 2 | ||||
| -rwxr-xr-x | examples/strains.scm | 19 |
9 files changed, 101 insertions, 175 deletions
diff --git a/examples/classification.scm b/examples/classification.scm index 279dc86..6195ea6 100755 --- a/examples/classification.scm +++ b/examples/classification.scm @@ -39,9 +39,7 @@ (gnc:species xkos:specializes gnc:set)) (triples "gnc:species" (set skos:member - (string->identifier "species" (remap-species-identifiers (field Species Fullname)) - #:separator "_" - #:proc string-downcase)))) + (string->identifier "" (remap-species-identifiers (field Species Fullname)))))) (define-transformer classification-scheme-set (tables (InbredSet)) @@ -53,8 +51,7 @@ (triples "gnc:set" (set skos:member (string->identifier - "set" (field InbredSet Name InbredSetName) - #:separator "_")))) + "set" (field InbredSet Name InbredSetName) #:separator "_")))) (define-transformer species (tables (Species)) @@ -69,9 +66,7 @@ (gnt:belongs_to_species rdf:comment "This resource given to this species") (gnt:belongs_to_species rdf:label "belongsToSpecies")) (triples - (string->identifier "species" (remap-species-identifiers (field Species Fullname)) - #:separator "_" - #:proc string-downcase) + (string->identifier "" (remap-species-identifiers (field Species Fullname))) (set skos:inScheme 'gnc:resource_classification_scheme) (set rdfs:label (remap-species-identifiers (field Species Fullname))) (set skos:prefLabel (field Species MenuName)) @@ -88,8 +83,8 @@ (left-join MappingMethod "ON InbredSet.MappingMethodId=MappingMethod.Id"))) (schema-triples - (gnt:genetic_type a owl:ObjectProperty) - (gnt:genetic_type rdfs:domain gnc:set) + (gnt:genetic-type a owl:ObjectProperty) + (gnt:genetic-type rdfs:domain gnc:set) (gnt:code a owl:ObjectProperty) (gnt:code rdfs:domain gnc:set) ;; Already defined as an owl prop in species @@ -98,21 +93,17 @@ (gnt:mapping_method rdfs:domain gnc:set) (gnt:belongs_to_group a rdf:property) (gnt:belongs_to_group rdf:comment "This resource given to this group") - (gnt:belongs_to_group rdf:label "belongs_to_group")) - (triples (string->identifier - "set" (field InbredSet Name InbredSetName) - #:separator "_") + (gnt:belongs_to_group rdf:label "belongs-to-group")) + (triples (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_") (set skos:inScheme 'gnc:resource_classification_scheme) (set rdfs:label (field InbredSet FullName)) (set skos:prefLabel (field InbredSet Name InbredSetName)) - (set gnt:genetic_type (field InbredSet GeneticType)) + (set gnt:genetic-type (field InbredSet GeneticType)) (set gnt:family (field InbredSet Family)) (set gnt:mapping_method (field MappingMethod Name)) (set gnt:code (field InbredSet InbredSetCode)) (set xkos:generalizes - (string->identifier "species" (remap-species-identifiers (field Species Fullname)) - #:separator "_" - #:proc string-downcase)))) + (string->identifier "" (remap-species-identifiers (field Species Fullname)))))) diff --git a/examples/dataset-metadata.scm b/examples/dataset-metadata.scm index 591b18e..c40c42c 100755 --- a/examples/dataset-metadata.scm +++ b/examples/dataset-metadata.scm @@ -27,7 +27,8 @@ (string->identifier "investigator" (string-join (list first-name last-name (fix-email-id email)) - "_"))) + "_") + #:separator "_")) (define-transformer investigators ;; There are a few duplicate entries. We group by email to @@ -36,7 +37,7 @@ "GROUP BY Email") (triples (investigator-attributes->id (field Investigators FirstName) (field Investigators LastName) - (field Investigators Email)) + "") (set rdf:type 'foaf:Person) (set foaf:name (string-append (field Investigators FirstName) " " (field Investigators LastName))) @@ -64,7 +65,7 @@ (gnt:has_go_tree_value a owl:ObjectProperty) (gnt:has_go_tree_value skos:definition "This resource the following GO tree value") (gnt:has_go_tree_value rdfs:domain gnc:gene_chip)) - (triples (string->identifier "platform" (field GeneChip Name)) + (triples (string->identifier "platform" (field GeneChip Name) #:separator "_") (set rdf:type 'gnc:gene_chip) (set rdfs:label (field GeneChip GeneChipName)) (set skos:prefLabel (field GeneChip Name)) @@ -72,9 +73,7 @@ Title))) (set gnt:has_go_tree_value (field GeneChip Go_tree_value)) (set xkos:classifiedUnder - (string->identifier "" (remap-species-identifiers (field Species Fullname)) - #:separator "" - #:proc string-capitalize-first)) + (string->identifier "" (remap-species-identifiers (field Species Fullname)) #:separator "")) (set gnt:has_geo_series_id (ontology 'geoSeries: (string-trim-both (field GeneChip GeoPlatform)))))) @@ -162,10 +161,15 @@ (gnt:has_acknowledgement rdfs:label "Acknowledgement") (gnt:has_acknowledgement a owl:ObjectProperty) (gnt:has_acknowledgement skos:definition "People to acknowledge")) - (triples (string->identifier - "" (regexp-substitute/global #f "[^A-Za-z0-9:]" - (field InfoFiles InfoPageName) - 'pre "_" 'post)) + (triples + (string->identifier + "" (let ((info-page-name (field InfoFiles InfoPageName)) + (info-title (field InfoFiles Title))) + (format #f "~a" + (if (and (string? info-page-name) + (string=? (string-downcase (string-trim-both info-page-name)) + "none")) + info-title info-page-name)))) (set rdf:type 'dcat:Dataset) (set xkos:classifiedUnder (let ([dataset-type @@ -176,30 +180,21 @@ (string->symbol dataset-type) ""))) - (set rdfs:label (regexp-substitute/global - #f "^[Nn]one$" - (field InfoFiles InfoPageName) - "")) + (set rdfs:label (normalize-string-field (field InfoFiles InfoPageName))) (set skos:prefLabel - (field ("IFNULL(GenoFreeze.FullName, IFNULL(PublishFreeze.FullName, ''))" - DatasetFullName))) + (normalize-string-field + (field ("IFNULL(GenoFreeze.FullName, IFNULL(PublishFreeze.FullName, ''))" + DatasetFullName)))) (set skos:altLabel (field Datasets DatasetName DatasetGroup)) - (set dct:title - (regexp-substitute/global - #f "^[Nn]one$" - (or - (regexp-substitute/global - #f "^Unpublished$" (field Datasets PublicationTitle) "") - (field InfoFiles InfoFileTitle) - "") - "")) + (set dct:title (normalize-string-field (field Datasets PublicationTitle))) (set dct:created - (field ("IFNULL(GenoFreeze.CreateTime, IFNULL(PublishFreeze.CreateTime, IFNULL(ProbeSetFreeze.CreateTime, '')))" - createTimeGenoFreeze))) + (normalize-string-field + (field ("IFNULL(GenoFreeze.CreateTime, IFNULL(PublishFreeze.CreateTime, IFNULL(ProbeSetFreeze.CreateTime, '')))" + createTimeGenoFreeze)))) (set dcat:contactPoint (investigator-attributes->id (field Investigators FirstName) (field Investigators LastName) - (field Investigators Email))) + "")) (set foaf:Organization (field Organizations OrganizationName)) (set dct:identifier (format #f "GN~a" (field InfoFiles GN_AccesionId))) @@ -209,14 +204,16 @@ (string->identifier "set" (field ("IFNULL(InbredSet.Name, IFNULL(PublishInbredSet.Name, GenoInbredSet.Name))" - InbredSetName)))) + InbredSetName)) + #:separator "_")) (set gnt:has_tissue (string->identifier "tissue" - (field Tissue Short_Name))) + (field Tissue Short_Name) + #:separator "_")) (set gnt:uses_normalization - (string->identifier "avgMethod" - ;; If AvgMethodName is NULL, assume N/A. - (if (string-blank? (field AvgMethod Name AvgMethodName)) - "N/A" (field AvgMethod Name AvgMethodName)))) + (let ((avg-method (normalize-string-field (field AvgMethod Name AvgMethodName)))) + (if (not (string-blank? avg-method)) + (string->identifier "avg_method" avg-method #:separator "_") + ""))) (set gnt:has_summary (let* ((summary-link (format @@ -375,7 +372,8 @@ "" (string->symbol acknowledgment-link)))) (set gnt:uses_platform (string->identifier "platform" - (field GeneChip Name GeneChip))) + (field GeneChip Name GeneChip) + #:separator "_")) (set gnt:has_geo_series_id (let ((s (string-match "GSE[0-9]*" @@ -391,11 +389,7 @@ (left-join InbredSet "ON PublishFreeze.InbredSetId = InbredSet.InbredSetId")) "WHERE PublishFreeze.public > 0 AND PublishFreeze.confidentiality < 1 AND InfoFiles.InfoFileId IS NULL") (triples - (string->identifier - "" - (regexp-substitute/global #f "[^A-Za-z0-9:]" - (field PublishFreeze Name) - 'pre "_" 'post)) + (string->identifier "" (field PublishFreeze Name)) (set rdf:type 'dcat:Dataset) (set xkos:classifiedUnder 'gnc:phenotype) (set dct:title (field PublishFreeze FullName)) @@ -407,8 +401,7 @@ (set gnt:belongs_to_group (string->identifier "set" (field InbredSet Name InbredSetName) - #:separator "_" - #:proc (lambda (x) x))))) + #:separator "_")))) (define-transformer genofreeze (tables (GenoFreeze @@ -416,15 +409,7 @@ (left-join InbredSet "ON GenoFreeze.InbredSetId = InbredSet.InbredSetId")) "WHERE GenoFreeze.public > 0 AND GenoFreeze.confidentiality < 1 AND InfoFiles.InfoPageName IS NULL") (triples - (string->identifier - "" - (regexp-substitute/global - #f "[^A-Za-z0-9:]" - (regexp-substitute/global - #f "[^A-Za-z0-9:]" - (field GenoFreeze Name) - 'pre "_" 'post) - 'pre "_" 'post)) + (string->identifier "" (field GenoFreeze Name)) (set rdf:type 'dcat:Dataset) (set xkos:classifiedUnder 'gnc:genotype) (set rdfs:label (field GenoFreeze Name)) @@ -454,19 +439,14 @@ (gnt:uses_data_scale a owl:ObjectProperty) (gnt:uses_data_scale skos:definition "Thi data scale this resource uses")) (triples - (string->identifier - "" - (regexp-substitute/global - #f "[^A-Za-z0-9:]" - (field ProbeSetFreeze Name) - 'pre "_" 'post)) + (string->identifier "" (field ProbeSetFreeze Name)) (set rdf:type 'dcat:Dataset) (set xkos:classifiedUnder 'gnc:probeset) (set gnt:uses_normalization - (string->identifier "avgMethod" - ;; If AvgMethodName is NULL, assume N/A. - (if (string-blank? (field AvgMethod Name AvgMethodName)) - "N/A" (field AvgMethod Name AvgMethodName)))) + (let ((avg-method (field AvgMethod Name AvgMethodName))) + (if (string-blank? avg-method) + #f + avg-method))) (set dct:title (field ProbeSetFreeze FullName)) (set rdfs:label (field ProbeSetFreeze ShortName)) (set skos:prefLabel (field ProbeSetFreeze Name)) @@ -475,15 +455,8 @@ (field ProbeSetFreeze CreateTime) '^^xsd:datetime)) (set gnt:uses_data_scale (field ProbeSetFreeze DataScale)) - (set gnt:has_tissue - (string->identifier - "tissue" - (field Tissue Short_Name))) - (set gnt:belongs_to_group - (string->identifier - "set" (field InbredSet Name InbredSetName) - #:separator "" - #:proc string-capitalize-first)))) + (set gnt:has_tissue (string->identifier "tissue" (field Tissue Short_Name) #:separator "_")) + (set gnt:belongs_to_group (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_")))) diff --git a/examples/genbank.scm b/examples/genbank.scm index 7aae5ba..993c3bf 100755 --- a/examples/genbank.scm +++ b/examples/genbank.scm @@ -25,9 +25,7 @@ (field Genbank Id)) (set gnt:has_sequence (field Genbank Sequence)) (set gnt:belongs_to_species - (string->identifier "" (remap-species-identifiers (field Species Fullname)) - #:separator "_" - #:proc string-downcase)))) + (string->identifier "" (remap-species-identifiers (field Species Fullname)))))) diff --git a/examples/genelist.scm b/examples/genelist.scm index 18fd30b..dbca921 100755 --- a/examples/genelist.scm +++ b/examples/genelist.scm @@ -77,12 +77,9 @@ (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)) + "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))) @@ -225,20 +222,14 @@ ""))) (set gnt:chromosome (field GeneList Chromosome)) (set gnt:tx_start (annotate-field - (field GeneList tx_start) + (field GeneList TxStart) '^^xsd:double)) (set gnt:tx_end (annotate-field - (field GeneList tx_end) + (field GeneList TxEnd) '^^xsd:double)) (set gnt:strand (string-trim-both (field GeneList Strand))) (set - gnt:belongs_to_species - (string->identifier - "" - (remap-species-identifiers - (string-trim-both (field Species Name))) - #:separator "" - #:proc string-capitalize-first)) + gnt:belongs_to_species (string->identifier "" (remap-species-identifiers (field Species Fullname)))) (set gnt:transcript (ontology 'transcript: @@ -259,10 +250,11 @@ (if (number? gene-uid) (number->string gene-uid) - gene-uid))) + gene-uid) + #:separator "_")) (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:gene_symbol (string-trim-both (field GeneList_rn33 geneSymbol))) (set gnt:chromosome (field GeneList_rn33 chromosome)) (set gnt:tx_start (annotate-field (field GeneList_rn33 txStart) @@ -280,7 +272,7 @@ gnc:has_kg_id (string-trim-both (field GeneList_rn33 kgID))) (set dct:references - (let ((symbol (field GeneList_rn33 gene_symbol))) + (let ((symbol (field GeneList_rn33 geneSymbol))) (if (not (string-blank? symbol)) (string->symbol (format #f @@ -290,7 +282,7 @@ "a gnc:PantherLink")) ""))) (set dct:references - (let ((symbol (string-trim-both (field GeneList_rn33 gene_symbol)))) + (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol)))) (if (not (string-blank? symbol)) (string->symbol (format #f @@ -300,7 +292,7 @@ "a gnc:ebi_gwas_link")) ""))) (set dct:references - (let ((symbol (string-trim-both (field GeneList_rn33 gene_symbol)))) + (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol)))) (if (not (string-blank? symbol)) (string->symbol (format #f @@ -311,7 +303,7 @@ "a gnc:panther_link")) ""))) (set dct:references - (let ((symbol (string-trim-both (field GeneList_rn33 gene_symbol)))) + (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol)))) (if (not (string-blank? symbol)) (string->symbol (format #f @@ -322,7 +314,7 @@ "a gnc:gtex_link")) ""))) (set dct:references - (let ((symbol (string-trim-both (field GeneList_rn33 gene_symbol)))) + (let ((symbol (string-trim-both (field GeneList_rn33 geneSymbol)))) (if (not (string-blank? symbol)) (string->symbol (format #f diff --git a/examples/generif.scm b/examples/generif.scm index 5fb95f7..994c540 100755 --- a/examples/generif.scm +++ b/examples/generif.scm @@ -24,33 +24,30 @@ GROUP BY GeneRIF.Id, GeneRIF.versionId, GeneRIF.symbol") (schema-triples (gnc:gene_wiki_entry a rdfs:Class) - (gnc:GNWikiEntry rdfs:subClassOf gnc:gene_wiki_entry) + (gnc:gn_wiki_entry rdfs:subClassOf gnc:gene_wiki_entry) (gnt:initial a owl:ObjectProperty) (gnt:initial rdfs:domain gnc:gene_wiki_entry) (gnt:initial skos:definition "Optional user or project code or your initials") (gnt:reason a owl:ObjectProperty) (gnt:reason rdfs:domain gnc:gene_wiki_entry) (gnt:reason skos:definition "The reason why this resource was modified") - (gnc:GNWikiEntry rdfs:comment "Represents GeneRIF Entries entered from GeneNetwork") - (gnt:gene_symbol rdfs:domain gnc:GNWikiEntry)) + (gnc:gn_wiki_entry rdfs:comment "Represents GeneRIF Entries entered from GeneNetwork") + (gnt:gene_symbol rdfs:domain gnc:gn_wiki_entry)) (triples - (format - #f "gn:wiki-~a-~a" - (field GeneRIF Id) - (field GeneRIF versionId)) + (string->identifier + "wiki" (format #f "~a_~a" + (field GeneRIF Id) + (field GeneRIF versionId)) + #:separator "_") (set rdfs:label (string->symbol (format #f "'~a'@en" (replace-substrings (sanitize-rdf-string (field GeneRIF comment)) '(("'" . "\\'")))))) - (set rdf:type 'gnc:GNWikiEntry) + (set rdf:type 'gnc:gn_wiki_entry) (set gnt:symbol (field GeneRIF symbol)) - (set gnt:belongs_to_species (string->identifier - "" - (remap-species-identifiers (field Species Fullname)) - #:separator "" - #:proc string-capitalize-first)) + (set gnt:belongs_to_species (string->identifier "" (remap-species-identifiers (field Species Fullname)))) (set dct:created (string->symbol (format #f "~s^^xsd:datetime " @@ -80,7 +77,7 @@ GROUP BY GeneRIF.Id, GeneRIF.versionId, GeneRIF.symbol") '^^xsd:integer)) (set gnt:initial (sanitize-rdf-string (field GeneRIF initial))) (set gnt:reason (field GeneRIF reason)) - (multiset gnt:belongsToCategory + (multiset gnt:belongs_to_category (string-split (field ("GROUP_CONCAT(DISTINCT GeneCategory.Name SEPARATOR ';')" GeneCategory)) @@ -93,13 +90,13 @@ GROUP BY GeneRIF.Id, GeneRIF.versionId, GeneRIF.symbol") (gnc:ncbi_wiki_entry rdfs:subClassOf gnc:gene_wiki_entry) (gnc:ncbi_wiki_entry rdfs:comment "Represents GeneRIF Entries obtained from NCBI")) (triples - (format - #f "gn:rif-~a-~a-~a-~a" - (field GeneRIF_BASIC GeneId) - (field GeneRIF_BASIC PubMed_ID) - (field - ("DATE_FORMAT(createtime, '%Y-%m-%dT%T')" CreateTime)) - (field GeneRIF_BASIC VersionId)) + (string->identifier + "rif" (format #f "~a_~a_~a_~a" + (field GeneRIF_BASIC GeneId) + (field GeneRIF_BASIC PubMed_ID) + (field ("DATE_FORMAT(createtime, '%Y-%m-%dT%T')" CreateTime)) + (field GeneRIF_BASIC VersionId)) + #:separator "_") (set rdf:type (let* ((comment (format #f "'~a'@en" (replace-substrings @@ -113,11 +110,7 @@ GROUP BY GeneRIF.Id, GeneRIF.versionId, GeneRIF.symbol") (field ("CAST(createtime AS CHAR)" EntryCreateTime)))) (symbol (field GeneRIF_BASIC symbol)) - (species (string->identifier - "" - (remap-species-identifiers (field Species Fullname)) - #:separator "" - #:proc string-capitalize-first)) + (species (string->identifier "" (remap-species-identifiers (field Species Fullname)))) (gene-id (field GeneRIF_BASIC GeneId)) (taxon-id (field GeneRIF_BASIC TaxID TaxonomicId)) (pmid (field GeneRIF_BASIC PubMed_ID)) @@ -172,8 +165,9 @@ GROUP BY GeneRIF.Id, GeneRIF.versionId, GeneRIF.symbol") ("owl:" "<http://www.w3.org/2002/07/owl#>"))) (inputs (list - gn-genewiki-entries - ncbi-genewiki-entries)) + ;; gn-genewiki-entries + ncbi-genewiki-entries + )) (outputs `(#:documentation ,documentation #:rdf ,output)))) diff --git a/examples/genotype.scm b/examples/genotype.scm index 257a3fa..0b0cd8e 100755 --- a/examples/genotype.scm +++ b/examples/genotype.scm @@ -46,14 +46,7 @@ (gnt:chr_num rdfs:domain gnc:genotype) (gnt:chr_num skos:definition "The chromosome number for this resource")) (triples - (string->identifier - "" - (regexp-substitute/global - #f "[^A-Za-z0-9:]" - (field Geno Name) - 'pre "_" 'post) - #:separator "_" - #:proc (lambda (x) x)) + (string->identifier "" (field Geno Name)) (set rdf:type 'gnc:genotype) (set rdfs:label (sanitize-rdf-string (field Geno Name))) (set gnt:chr (field Geno Chr)) diff --git a/examples/phenotype.scm b/examples/phenotype.scm index 1bec264..5a118b8 100755 --- a/examples/phenotype.scm +++ b/examples/phenotype.scm @@ -56,14 +56,12 @@ "trait" (field ("CONCAT(IFNULL(InbredSet.InbredSetCode, PublishXRef.InbredSetId), '_', PublishXRef.Id)" Phenotype)) - #:separator "_" - #:proc (lambda (x) x)) + #:separator "_") (set rdf:type 'gnc:phenotype) (set gnt:belongs_to_group (string->identifier "set" (field InbredSet Name InbredSetName) - #:separator "_" - #:proc string-capitalize-first)) + #:separator "_")) ;; This is the trait's name (set gnt:trait_id (let ((trait-id (field PublishXRef Id))) @@ -85,15 +83,7 @@ (set dct:contributor (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:locus (sanitize-rdf-string (field PublishXRef Locus))) (set gnt:lod_score (annotate-field (field ("IFNULL((PublishXRef.LRS/4.604), '')" lrs)) '^^xsd:double)) diff --git a/examples/schema.scm b/examples/schema.scm index 50cfd6a..bfc5747 100755 --- a/examples/schema.scm +++ b/examples/schema.scm @@ -33,7 +33,7 @@ (table-name table))))) (triple table-id 'rdf:type 'gn:sqlTable) (triple table-id 'gn:name (table-name table)) - (triple table-id 'gn:hasSize (string->symbol (format #f "~a" (table-size table)))) + (triple table-id 'gn:has_size (string->symbol (format #f "~a" (table-size table)))) (for-each (lambda (column) (let ((column-id (column-id (table-name table) (column-name column)))) diff --git a/examples/strains.scm b/examples/strains.scm index 976d358..b2332bd 100755 --- a/examples/strains.scm +++ b/examples/strains.scm @@ -63,16 +63,11 @@ At this point it is not very clear how Name, Name2, Symbol and Alias are used. (gnt:gene_symbol rdfs:domain gnc:strain) (gnt:gene_symbol a owl:ObjectProperty)) (triples (string->identifier - "" - (regexp-substitute/global - #f "[^A-Za-z0-9:]" - (field Strain Name) - 'pre "_" 'post)) + "strain" + (field Strain Name) + #:separator "_") (set rdf:type 'gnc:strain) - (set gnt:belongs_to_species - (string->identifier "" (remap-species-identifiers (field Species Fullname)) - #:separator "_" - #:proc string-downcase)) + (set gnt:belongs_to_species (string->identifier "" (remap-species-identifiers (field Species Fullname)))) ;; Name, and maybe a second name (set rdfs:label (sanitize-rdf-string (field Strain Name))) (set skos:altLabel (sanitize-rdf-string (field ("IF ((Strain.Name2 != Strain.Name), Strain.Name2, '')" Name2)))) @@ -85,7 +80,7 @@ At this point it is not very clear how Name, Name2, Symbol and Alias are used. (gnc:mapping_method a skos:Concept) (gnc:mapping_method skos:definition "Terms that decribe mapping methods used on this resource")) (triples - (string->identifier "mapping_method" (field MappingMethod Name)) + (string->identifier "mapping_method" (field MappingMethod Name) #:separator "_") (set rdf:type 'gnc:mapping_method) (set rdfs:label (field MappingMethod Name)))) @@ -96,7 +91,7 @@ At this point it is not very clear how Name, Name2, Symbol and Alias are used. (schema-triples (gnc:avg_method a skos:Concept) (gnc:avg_method skos:definition "Terms that decribe normalization methods used on this resource")) - (triples (string->identifier "avgMethod" (field AvgMethod Name AvgMethodName)) + (triples (string->identifier "avg_method" (field AvgMethod Name AvgMethodName) #:separator "_") (set rdf:type 'gnc:avg_method) (set rdfs:label (field AvgMethod Normalization)))) @@ -115,7 +110,7 @@ At this point it is not very clear how Name, Name2, Symbol and Alias are used. read))) (with-documentation - (name "Species Metadata") + (name "Strain Metadata") (connection %connection-settings) (table-metadata? #f) (prefixes |