diff options
Diffstat (limited to 'examples/dump-phenotype.scm')
| -rwxr-xr-x | examples/dump-phenotype.scm | 76 |
1 files changed, 38 insertions, 38 deletions
diff --git a/examples/dump-phenotype.scm b/examples/dump-phenotype.scm index 1ef498d..00f99d2 100755 --- a/examples/dump-phenotype.scm +++ b/examples/dump-phenotype.scm @@ -27,10 +27,10 @@ (left-join InbredSet "ON PublishFreeze.InbredSetId = InbredSet.InbredSetId")) "WHERE PublishFreeze.public > 0 AND PublishFreeze.confidentiality < 1 AND InfoFiles.InfoPageName IS NULL") (schema-triples - (gn-term:datasetOfInbredSet rdfs:range gn:inbredSet) - (gn-term:name rdfs:range rdfs:Literal) - (gn-term:fullName rdfs:range rdfs:Literal) - (gn-term:shortName rdfs:range rdfs:Literal) + (gnt:datasetOfInbredSet rdfs:range gn:inbredSet) + (gnt:name rdfs:range rdfs:Literal) + (gnt:fullName rdfs:range rdfs:Literal) + (gnt:shortName rdfs:range rdfs:Literal) (gn:phenotypeDataset rdf:subClassOf gn:dataset)) (triples (string->identifier @@ -41,13 +41,13 @@ #:separator "" #:proc string-capitalize-first) (set rdf:type 'gn:phenotypeDataset) - (set gn-term:name (field PublishFreeze Name)) - (set gn-term:fullName (field PublishFreeze FullName)) - (set gn-term:shortName (field PublishFreeze ShortName)) + (set gnt:name (field PublishFreeze Name)) + (set gnt:fullName (field PublishFreeze FullName)) + (set gnt:shortName (field PublishFreeze ShortName)) (set dct:created (annotate-field (field PublishFreeze CreateTime) '^^xsd:date)) - (set gn-term:datasetOfInbredSet + (set gnt:datasetOfInbredSet (string->identifier "inbredSet" (field InbredSet Name InbredSetName))))) (define-dump dump-phenotypes @@ -58,19 +58,19 @@ (left-join InfoFiles "ON InfoFiles.InfoPageName = PublishFreeze.Name"))) (schema-triples (gn:phenotypeDataset rdfs:subPropertyOf gn:dataset) - (gn-term:publicationDescription rdfs:range rdfs:Literal) - (gn-term:originalDescription rdfs:range rdfs:Literal) - (gn-term:prePublicationDescription rdfs:range rdfs:Literal) - (gn-term:postPublicationAbbreviation rdfs:range rdfs:Literal) - (gn-term:labCode rdfs:range rdfs:Literal) - (gn-term:submitter rdfs:range rdfs:Literal) - (gn-term:owner rdfs:range rdfs:Literal) - (gn-term:mean rdfs:range xsd:double) - (gn-term:LRS rdfs:range xsd:float) - (gn-term:locus rdfs:range rdfs:Literal) - (gn-term:additive rdfs:range xsd:decimal) - (gn-term:sequence rdfs:range rdfs:Literal) - (gn-term:phenotypeOfPublication rdfs:range gn-term:pubMedId)) + (gnt:publicationDescription rdfs:range rdfs:Literal) + (gnt:originalDescription rdfs:range rdfs:Literal) + (gnt:prePublicationDescription rdfs:range rdfs:Literal) + (gnt:postPublicationAbbreviation rdfs:range rdfs:Literal) + (gnt:labCode rdfs:range rdfs:Literal) + (gnt:submitter rdfs:range rdfs:Literal) + (gnt:owner rdfs:range rdfs:Literal) + (gnt:mean rdfs:range xsd:double) + (gnt:LRS rdfs:range xsd:float) + (gnt:locus rdfs:range rdfs:Literal) + (gnt:additive rdfs:range xsd:decimal) + (gnt:sequence rdfs:range rdfs:Literal) + (gnt:phenotypeOfPublication rdfs:range gn-term:pubMedId)) (triples (string->identifier "" (regexp-substitute/global #f "[^A-Za-z0-9:]" @@ -79,44 +79,44 @@ #:separator "" #:proc string-capitalize-first) (set rdf:type 'gn:phenotype) - (set gn-term:name (sanitize-rdf-string + (set gnt:name (sanitize-rdf-string (field ("CAST(CONVERT(BINARY CONVERT(IF(Phenotype.Post_publication_abbreviation IS NULL, IF(Phenotype.Pre_publication_abbreviation IS NULL, Phenotype.Id, Phenotype.Pre_publication_abbreviation), Phenotype.Post_publication_abbreviation) USING latin1) USING utf8) AS VARCHAR(100))" PhenotypeName)))) ;; There is no row with an empty post-publication description so ;; use this field as the main publication description - (set gn-term:publicationDescription + (set gnt:publicationDescription (sanitize-rdf-string (field ("CAST(CONVERT(BINARY CONVERT(Phenotype.Post_publication_description USING latin1) USING utf8) AS CHAR(10000))" postPubDescr)))) - (set gn-term:originalDescription (sanitize-rdf-string + (set gnt:originalDescription (sanitize-rdf-string (delete-substrings (field Phenotype Original_description) "Original post publication description: "))) - (set gn-term:prePublicationDescription + (set gnt:prePublicationDescription (sanitize-rdf-string (field ("CAST(CONVERT(BINARY CONVERT(Phenotype.Pre_publication_description USING latin1) USING utf8) AS VARCHAR(15000))" prePubDesc)))) - (set gn-term:prePublicationAbbreviation (sanitize-rdf-string (field Phenotype Pre_publication_abbreviation))) - (set gn-term:postPublicationAbbreviation (sanitize-rdf-string (field Phenotype Post_publication_abbreviation))) - (set gn-term:labCode (field Phenotype Lab_code)) - (set gn-term:submitter (sanitize-rdf-string (field Phenotype Submitter))) - (set gn-term:owner (sanitize-rdf-string (field Phenotype Owner))) - (set gn-term:mean (annotate-field (field ("IFNULL(PublishXRef.mean, '')" mean)) + (set gnt:prePublicationAbbreviation (sanitize-rdf-string (field Phenotype Pre_publication_abbreviation))) + (set gnt:postPublicationAbbreviation (sanitize-rdf-string (field Phenotype Post_publication_abbreviation))) + (set gnt:labCode (field Phenotype Lab_code)) + (set gnt:submitter (sanitize-rdf-string (field Phenotype Submitter))) + (set gnt:owner (sanitize-rdf-string (field Phenotype Owner))) + (set gnt:mean (annotate-field (field ("IFNULL(PublishXRef.mean, '')" mean)) '^^xsd:double)) - (set gn-term:locus (field PublishXRef Locus)) - (set gn-term:LRS (annotate-field (field ("IFNULL(PublishXRef.LRS, '')" lrs)) '^^xsd:float)) - (set gn-term:additive (annotate-field (field ("IFNULL(PublishXRef.additive, '')" additive)) '^^xsd:decimal)) - (set gn-term:sequence (annotate-field (field PublishXRef Sequence) '^^xsd:int)) - (set gn-term:phenotypeOfDataset + (set gnt:locus (field PublishXRef Locus)) + (set gnt:LRS (annotate-field (field ("IFNULL(PublishXRef.LRS, '')" lrs)) '^^xsd:float)) + (set gnt:additive (annotate-field (field ("IFNULL(PublishXRef.additive, '')" additive)) '^^xsd:decimal)) + (set gnt:sequence (annotate-field (field PublishXRef Sequence) '^^xsd:int)) + (set gnt:phenotypeOfDataset (string->identifier "" (field ("IFNULL(InfoFiles.InfoPageName, IFNULL(PublishFreeze.Name, ''))" DatasetName)) #:separator "" #:proc string-capitalize-first)) - (set gn-term:phenotypeOfPublication + (set gnt:phenotypeOfPublication (let ((pmid (field ("IF(Publication.PubMed_ID IS NULL, '', CONVERT(Publication.PubMed_Id, INT))" pmid))) @@ -134,7 +134,7 @@ (prefixes '(("dct:" "<http://purl.org/dc/terms/>") ("gn:" "<http://genenetwork.org/id/>") - ("gn-term:" "<http://genenetwork.org/terms/>") + ("gnt:" "<http://genenetwork.org/terms/>") ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>") ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>") ("xsd:" "<http://www.w3.org/2001/XMLSchema#>") |