diff options
Diffstat (limited to 'examples')
| -rwxr-xr-x | examples/dataset-metadata.scm | 176 | ||||
| -rwxr-xr-x | examples/genbank.scm | 3 | ||||
| -rwxr-xr-x | examples/genelist.scm | 2 | ||||
| -rwxr-xr-x | examples/generif-old.scm | 241 | ||||
| -rwxr-xr-x | examples/generif.scm | 255 | ||||
| -rwxr-xr-x | examples/probeset.scm | 203 | ||||
| -rwxr-xr-x | examples/schema.scm | 70 |
7 files changed, 594 insertions, 356 deletions
diff --git a/examples/dataset-metadata.scm b/examples/dataset-metadata.scm index 5c59530..9c30180 100755 --- a/examples/dataset-metadata.scm +++ b/examples/dataset-metadata.scm @@ -146,6 +146,26 @@ (gnt:hasCaseInfo rdfs:label "About Case") (gnt:hasCaseInfo a owl:ObjectProperty) (gnt:hasCaseInfo skos:definition "Information about the cases used in this platform") + (gnt:hasSummary rdfs:domain dcat:Dataset) + (gnt:hasSummary rdfs:label "Summary") + (gnt:hasSummary a owl:ObjectProperty) + (gnt:hasSummary skos:definition "Summary information about dataset") + (gnt:hasCitation rdfs:domain dcat:Dataset) + (gnt:hasCitation rdfs:label "Citation") + (gnt:hasCitation a owl:ObjectProperty) + (gnt:hasCitation skos:definition "Citation for this dataset") + (gnt:hasContributors rdfs:domain dcat:Dataset) + (gnt:hasContributors rdfs:label "Contributors") + (gnt:hasContributors a owl:ObjectProperty) + (gnt:hasContributors skos:definition "Contributors of this resource") + (gnt:hashasExperimentDesign rdfs:domain dcat:Dataset) + (gnt:hashasExperimentDesign rdfs:label "Experiment Design") + (gnt:hashasExperimentDesign a owl:ObjectProperty) + (gnt:hashasExperimentDesign skos:definition "Experiment Design for this resource") + (gnt:hasTissueInfo rdfs:domain dcat:Dataset) + (gnt:hasTissueInfo rdfs:label "Tissue Information") + (gnt:hasTissueInfo a owl:ObjectProperty) + (gnt:hasTissueInfo skos:definition "Tissue information about dataset") (gnt:hasExperimentType skos:definition "Information about the experiment type") (gnt:hasAcknowledgement rdfs:domain dcat:Dataset) (gnt:hasAcknowledgement rdfs:label "Acknowledgement") @@ -206,6 +226,162 @@ ;; If AvgMethodName is NULL, assume N/A. (if (string-blank? (field AvgMethod Name AvgMethodName)) "N/A" (field AvgMethod Name AvgMethodName)))) + (set gnt:hasSummary + (let* ((summary-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/summary.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (summary + (field InfoFiles Summary))) + (if (or (null? summary) (string-blank? summary)) + "" (string->symbol summary-link)))) + (set gnt:hasTissueInfo + (let* ((tissue-info-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/tissue.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (tissue-info + (field Datasets AboutTissue))) + (if (or (null? tissue-info) (string-blank? tissue-info)) + "" (string->symbol tissue-info-link)))) + (set gnt:hasCitation + (let* ((citation-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/citation.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (citation + (field Datasets Citation))) + (if (or (null? citation) (string-blank? citation)) + "" (string->symbol citation-link)))) + (set gnt:hasSpecifics + (let* ((specifics-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/specifics.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (specifics + (field InfoFiles Specifics))) + (if (or (null? specifics) (string-blank? specifics)) + "" (string->symbol specifics-link)))) + (set gnt:hasCaseInfo + (let* ((cases-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/cases.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (cases + (field Datasets AboutCases))) + (if (or (null? cases) (string-blank? cases)) + "" (string->symbol cases-link)))) + (set gnt:hasPlatformInfo + (let* ((platform-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/platform.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (platform + (field Datasets AboutPlatform))) + (if (or (null? platform) (string-blank? platform)) + "" (string->symbol platform-link)))) + (set gnt:hasDataProcessingInfo + (let* ((processing-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/processing.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (processing + (field Datasets AboutDataProcessing))) + (if (or (null? processing) (string-blank? processing)) + "" (string->symbol processing-link)))) + (set gnt:hasNotes + (let* ((notes-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/notes.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (notes + (field Datasets Notes))) + (if (or (null? notes) (string-blank? notes)) + "" (string->symbol notes-link)))) + (set gnt:hasExperimentType + (let* ((experiment-type-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/experiment-type.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (experiment-type + (field InfoFiles Experiment_Type))) + (if (or (null? experiment-type) (string-blank? experiment-type)) + "" (string->symbol experiment-type-link)))) + (set gnt:hasExperimentDesign + (let* ((experiment-design-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/experiment-design.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (experiment-design + (field Datasets ExperimentDesign))) + (if (or (null? experiment-design) (string-blank? experiment-design)) + "" (string->symbol experiment-design-link)))) + (set gnt:hasContributors + (let* ((contributors-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/contributors.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (contributors + (field Datasets Contributors))) + (if (or (null? contributors) (string-blank? contributors)) + "" (string->symbol contributors-link)))) + (set gnt:hasAcknowledgement + (let* ((acknowledgment-link + (format + #f "<https://git.genenetwork.org/gn-docs/tree/general/datasets/~a/acknowledgment.rtf>" + (string-capitalize-first + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field InfoFiles InfoPageName) + 'pre "_" 'post)))) + (acknowledgment + (field Datasets Acknowledgment))) + (if (or (null? acknowledgment) (string-blank? acknowledgment)) + "" (string->symbol acknowledgment-link)))) (set gnt:usesPlatform (string->identifier "platform" (field GeneChip Name GeneChip))) diff --git a/examples/genbank.scm b/examples/genbank.scm index 391cff0..c83643c 100755 --- a/examples/genbank.scm +++ b/examples/genbank.scm @@ -10,8 +10,7 @@ (transform strings) (transform sql) (transform triples) - (transform special-forms) - (transform uuid)) + (transform special-forms)) diff --git a/examples/genelist.scm b/examples/genelist.scm index 9c1ced0..8729626 100755 --- a/examples/genelist.scm +++ b/examples/genelist.scm @@ -18,6 +18,8 @@ (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) diff --git a/examples/generif-old.scm b/examples/generif-old.scm new file mode 100755 index 0000000..ede5a28 --- /dev/null +++ b/examples/generif-old.scm @@ -0,0 +1,241 @@ +#! /usr/bin/env guile +!# + +(use-modules (srfi srfi-1) + (srfi srfi-26) + (rnrs bytevectors) + (ice-9 format) + (ice-9 getopt-long) + (ice-9 match) + (ice-9 regex) + (transform strings) + (transform sql) + (transform triples) + (transform special-forms)) + + + +(define (fix-email-id email) + (string-delete #\space email)) + +(define (investigator-attributes->id first-name last-name email) + ;; There is just one record corresponding to "Evan Williams" which + ;; does not have an email ID. To accommodate that record, we + ;; construct the investigator ID from not just the email ID, but + ;; also the first and the last names. It would be preferable to just + ;; find Evan Williams' email ID and insert it into the database. + (string->identifier "investigator" + (string-join + (list first-name last-name (fix-email-id email)) + "_"))) + + + +(define-transformer genewiki-symbols + (tables (GeneRIF_BASIC) + "GROUP BY BINARY symbol") + (triples + (string->identifier + "symbol" + (regexp-substitute/global #f "[^A-Za-z0-9:]" + (field GeneRIF_BASIC symbol) + 'pre "_" 'post) + #:proc (lambda (x) x)) + (set rdfs:label + (field GeneRIF_BASIC symbol)))) + +;; Some symbols exist in the RIF table that don't exist in the GeneRIF +;; table. +(define-transformer generif-symbols + (tables (GeneRIF) + "WHERE symbol NOT IN (SELECT symbol from GeneRIF_BASIC) GROUP BY BINARY symbol") + (triples + (string->identifier + "symbol" + (regexp-substitute/global #f "[^A-Za-z0-9:]" + (field GeneRIF symbol) + 'pre "_" 'post) + #:proc (lambda (x) x)) + (set rdfs:label + (field GeneRIF symbol)))) + +(define-transformer gn-genewiki-entries + (tables (GeneRIF + (left-join Species "ON Species.SpeciesId = GeneRIF.SpeciesId") + (left-join GeneRIFXRef "ON GeneRIFXRef.GeneRIFId = GeneRIF.Id") + (left-join GeneCategory "ON GeneRIFXRef.GeneCategoryId = GeneCategory.Id") + (left-join Investigators "ON Investigators.Email = GeneRIF.email")) + "WHERE GeneRIF.display > 0 AND GeneRIF.VersionId = 0 AND GeneRIF.comment IS NOT NULL GROUP BY GeneRIF.comment, BINARY GeneRIF.symbol") + (schema-triples + (gnc:GeneWikiEntry a rdfs:Class) + (gnc:GNWikiEntry rdfs:subClassOf gnc:GeneWikiEntry) + (gnc:GNWikiEntry rdfs:comment "Represents GeneRIF Entries entered from GeneNetwork") + (gnt:geneSymbol rdfs:domain gnc:GNWikiEntry)) + (triples + (string->identifier + "symbol" + (regexp-substitute/global + #f "[^A-Za-z0-9:]" + (field GeneRIF symbol) + 'pre "_" 'post) + #:proc (lambda (x) x)) + (set rdfs:comment + (let* ([generif-comment (sanitize-rdf-string (field GeneRIF comment))] + [create-time (field GeneRIF createtime EntryCreateTime)] + [pmid (field GeneRIF PubMed_ID PMID)] + [web-url (field GeneRIF weburl)] + [species (string->identifier + "" + (remap-species-identifiers (field Species Fullname)) + #:separator "" + #:proc string-capitalize-first)] + [categories + (remove (lambda (x) + (or (eq? x #f) + (and (string? x) + (string-null? x)))) + (remove-duplicates + (string-split-substring + (field ("GROUP_CONCAT(DISTINCT GeneCategory.Name SEPARATOR '$$')" + GeneCategory)) + "$$")))]) + (string->symbol + (string-append + "[ " + (format #f "rdf:type gnc:GNWikiEntry ; ") + (if (string? species) + "" + (format #f "gnt:belongsToSpecies ~a ; " + species)) + (format #f "rdfs:comment ~s^^xsd:string ; " + generif-comment) + (if (string? create-time) + "" + (format #f "dct:created ~s^^xsd:datetime ; " + (time-unix->string + create-time "~5"))) + (if (and (string? pmid) (not (string-null? pmid))) + (format #f + "~{dct:references pubmed:~a ; ~}" + (string-split pmid #\space)) + "") + (if (and (not (string-null? + (string-trim-both (field GeneRIF email)))) + (not (string-null? (field Investigators Email)))) + (format #f "dct:creator ~a ; " + (investigator-attributes->id + (field Investigators FirstName) + (field Investigators LastName) + (field Investigators Email))) + "") + (if (not (null? categories)) + (format #f + "~{gnt:belongsToCategory ~s ; ~}" + categories) + "") + (if (and (string? web-url) (not (string-null? web-url))) + (format #f "foaf:homepage ~s ; " + web-url) + "") + " ] ")))))) + +(define-transformer ncbi-genewiki-entries + (tables (GeneRIF_BASIC + (left-join Species "USING (SpeciesId)")) + "WHERE GeneRIF_BASIC.comment IS NOT NULL AND TRIM(GeneRIF_BASIC.comment) != '' AND TRIM(GeneRIF_BASIC.symbol) != '' GROUP BY GeneRIF_BASIC.comment, GeneRIF_BASIC.createtime, GeneRIF_BASIC.VersionId, GeneRIF_BASIC.SpeciesId, GeneRIF_BASIC.TaxID") + (schema-triples + (gnc:NCBIWikiEntry rdfs:subClassOf gnc:GeneWikiEntry) + (gnc:NCBIWikiEntry rdfs:comment "Represents GeneRIF Entries obtained from NCBI") + (gnt:hasVersionId a owl:ObjectProperty) + (gnt:hasVersionId rdfs:domain gnc:NCBIWikiEntry) + (gnt:hasVersionId skos:definition "The VersionId of this this resource")) + (triples + (string->identifier + "symbol" + (regexp-substitute/global #f "[^A-Za-z0-9:]" + (field GeneRIF_BASIC symbol GeneRIFSymbol) + 'pre "_" 'post) + #:proc (lambda (x) x)) + (set rdfs:comment + (let ([ncbi-comment (sanitize-rdf-string (field GeneRIF_BASIC comment))] + [species-name + (string->identifier + "" + (remap-species-identifiers (field Species Fullname SpeciesFullName)) + #:separator "" + #:proc string-capitalize-first)] + [taxonomic-id (field GeneRIF_BASIC TaxID TaxonomicId)] + [create-time (field GeneRIF_BASIC createtime EntryCreateTime)] + [pmid (field GeneRIF_BASIC PubMed_ID PMID)] + [gene-id (field GeneRIF_BASIC GeneId)] + [version-id (field GeneRIF_BASIC VersionId)]) + (string->symbol + (string-append + "[ " + (format #f "rdf:type gnc:NCBIWikiEntry ; ") + (format #f "rdfs:comment ~s^^xsd:string ; " + ncbi-comment) + (format #f "gnt:belongsToSpecies ~a ; " + species-name) + (if (eq? #f taxonomic-id) + "" + (format #f "skos:notation taxon:~a ; " + taxonomic-id)) + (format #f "gnt:hasGeneId generif:~a ; " + gene-id) + (format #f "gnt:hasVersionId '~a'^^xsd:integer ; " + version-id) + (if (and (string? pmid) (not (string-null? pmid))) + (format #f + "~{dct:references pubmed:~a ; ~}" + (string-split pmid #\space)) + "") + (if (string? create-time) + "" + (format #f "dct:created ~s^^xsd:datetime ; " + (time-unix->string + create-time "~5"))) + " ]")))))) + + + +(let* ((option-spec + '((settings (single-char #\s) (value #t)) + (output (single-char #\o) (value #t)) + (documentation (single-char #\d) (value #t)))) + (options (getopt-long (command-line) option-spec)) + (settings (option-ref options 'settings #f)) + (output (option-ref options 'output #f)) + (documentation (option-ref options 'documentation #f)) + (%connection-settings + (call-with-input-file settings + read))) + + (with-documentation + (name "GeneRIF Metadata") + (connection %connection-settings) + (table-metadata? #f) + (prefixes + '(("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>") + ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>") + ("skos:" "<http://www.w3.org/2004/02/skos/core#>") + ("xkos:" "<http://rdf-vocabulary.ddialliance.org/xkos#>") + ("gn:" "<http://genenetwork.org/id/>") + ("gnc:" "<http://genenetwork.org/category/>") + ("gnt:" "<http://genenetwork.org/term/>") + ("dct:" "<http://purl.org/dc/terms/>") + ("foaf:" "<http://xmlns.com/foaf/0.1/>") + ("pubmed:" "<http://rdf.ncbi.nlm.nih.gov/pubmed/>") + ("taxon:" "<https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mode=Info&id=>") + ("generif:" "<http://www.ncbi.nlm.nih.gov/gene?cmd=Retrieve&dopt=Graphics&list_uids=>") + ("xsd:" "<http://www.w3.org/2001/XMLSchema#>") + ("owl:" "<http://www.w3.org/2002/07/owl#>"))) + (inputs + (list + genewiki-symbols + generif-symbols + gn-genewiki-entries + ncbi-genewiki-entries)) + (outputs + `(#:documentation ,documentation + #:rdf ,output)))) diff --git a/examples/generif.scm b/examples/generif.scm index fb3208a..628e34e 100755 --- a/examples/generif.scm +++ b/examples/generif.scm @@ -11,177 +11,131 @@ (transform strings) (transform sql) (transform triples) - (transform special-forms) - (transform uuid)) + (transform special-forms)) -(define (fix-email-id email) - (string-delete #\space email)) - -(define (investigator-attributes->id first-name last-name email) - ;; There is just one record corresponding to "Evan Williams" which - ;; does not have an email ID. To accommodate that record, we - ;; construct the investigator ID from not just the email ID, but - ;; also the first and the last names. It would be preferable to just - ;; find Evan Williams' email ID and insert it into the database. - (string->identifier "investigator" - (string-join - (list first-name last-name (fix-email-id email)) - "_"))) - - - -(define-transformer genewiki-symbols - (tables (GeneRIF_BASIC) - "GROUP BY BINARY symbol") - (triples - (string->identifier - "symbol" - (regexp-substitute/global #f "[^A-Za-z0-9:]" - (field GeneRIF_BASIC symbol) - 'pre "_" 'post) - #:proc (lambda (x) x)) - (set rdfs:label - (field GeneRIF_BASIC symbol)))) - (define-transformer gn-genewiki-entries (tables (GeneRIF (left-join Species "ON Species.SpeciesId = GeneRIF.SpeciesId") (left-join GeneRIFXRef "ON GeneRIFXRef.GeneRIFId = GeneRIF.Id") - (left-join GeneCategory "ON GeneRIFXRef.GeneCategoryId = GeneCategory.Id") - (left-join Investigators "ON Investigators.Email = GeneRIF.email")) - "WHERE GeneRIF.display > 0 AND GeneRIF.VersionId = 0 AND GeneRIF.comment IS NOT NULL GROUP BY GeneRIF.comment, BINARY GeneRIF.symbol") + (left-join GeneCategory "ON GeneRIFXRef.GeneCategoryId = GeneCategory.Id")) + "WHERE GeneRIF.display > 0 AND GeneRIF.comment IS NOT NULL +GROUP BY GeneRIF.Id, GeneRIF.versionId, GeneRIF.symbol") (schema-triples (gnc:GeneWikiEntry a rdfs:Class) (gnc:GNWikiEntry rdfs:subClassOf gnc:GeneWikiEntry) + (gnt:initial a owl:ObjectProperty) + (gnt:initial rdfs:domain gnc:GeneWikiEntry) + (gnt:initial skos:definition "Optional user or project code or your initials") + (gnt:reason a owl:ObjectProperty) + (gnt:reason rdfs:domain gnc:GeneWikiEntry) + (gnt:reason skos:definition "The reason why this resource was modified") (gnc:GNWikiEntry rdfs:comment "Represents GeneRIF Entries entered from GeneNetwork") (gnt:geneSymbol rdfs:domain gnc:GNWikiEntry)) (triples - (string->identifier - "symbol" - (regexp-substitute/global - #f "[^A-Za-z0-9:]" - (field GeneRIF symbol) - 'pre "_" 'post) - #:proc (lambda (x) x)) - (set rdfs:comment - (let* ([generif-comment (sanitize-rdf-string (field GeneRIF comment))] - [create-time (field GeneRIF createtime EntryCreateTime)] - [pmid (field GeneRIF PubMed_ID PMID)] - [web-url (field GeneRIF weburl)] - [species (string->identifier - "" - (remap-species-identifiers (field Species Fullname)) - #:separator "" - #:proc string-capitalize-first)] - [categories - (remove (lambda (x) - (or (eq? x #f) - (and (string? x) - (string-null? x)))) - (remove-duplicates - (string-split-substring - (field ("GROUP_CONCAT(DISTINCT GeneCategory.Name SEPARATOR '$$')" - GeneCategory)) - "$$")))]) - (string->symbol - (string-append - "[ " - (format #f "rdf:type gnc:GNWikiEntry ; ") - (if (string? species) - "" - (format #f "gnt:belongsToSpecies ~a ; " - species)) - (format #f "rdfs:comment ~s^^xsd:string ; " - generif-comment) - (if (string? create-time) - "" - (format #f "dct:created ~s^^xsd:datetime ; " - (time-unix->string - create-time "~5"))) - (if (and (string? pmid) (not (string-null? pmid))) - (format #f - "~{dct:references pubmed:~a ; ~}" - (string-split pmid #\space)) - "") - (if (and (not (string-null? - (string-trim-both (field GeneRIF email)))) - (not (string-null? (field Investigators Email)))) - (format #f "dct:creator ~a ; " - (investigator-attributes->id - (field Investigators FirstName) - (field Investigators LastName) - (field Investigators Email))) - "") - (if (not (null? categories)) - (format #f - "~{gnt:belongsToCategory ~s ; ~}" - categories) - "") - (if (and (string? web-url) (not (string-null? web-url))) - (format #f "foaf:homepage ~s ; " - web-url) - "") - " ] ")))))) + (format + #f "gn:wiki-~a-~a" + (field GeneRIF Id) + (field GeneRIF versionId)) + (set rdfs:label (string->symbol + (format #f "'~a'@en" + (replace-substrings + (sanitize-rdf-string + (field GeneRIF comment)) + '(("'" . "\\'")))))) + (set rdf:type 'gnc:GNWikiEntry) + (set gnt:symbol (field GeneRIF symbol)) + (set gnt:belongsToSpecies (string->identifier + "" + (remap-species-identifiers (field Species Fullname)) + #:separator "" + #:proc string-capitalize-first)) + (set dct:created + (string->symbol + (format #f "~s^^xsd:datetime " + (field + ("CAST(createtime AS CHAR)" EntryCreateTime))))) + (multiset dct:references + (map (lambda (pmid) + (match pmid + ((? string-blank? p) "") + (p (string->symbol + (format #f "pubmed:~a" (string-trim-both pmid)))))) + (string-split (field GeneRIF PubMed_ID PMID) + #\space))) + (set foaf:mbox + (match (sanitize-rdf-string (field GeneRIF email)) + ((? string-blank? mbox) "") + (mbox (string->symbol + (format #f "<~a>" mbox))))) + (set dct:identifier (annotate-field (format #f "~s" (field GeneRIF Id)) + '^^xsd:integer)) + (set foaf:homepage + (match (sanitize-rdf-string (field GeneRIF weburl)) + ((? string-blank? homepage) "") + (homepage (string->symbol + (format #f "<~a>" homepage))))) + (set dct:hasVersion (annotate-field (format #f "~s" (field GeneRIF versionId)) + '^^xsd:integer)) + (set gnt:initial (sanitize-rdf-string (field GeneRIF initial))) + (set gnt:reason (field GeneRIF reason)) + (multiset gnt:belongsToCategory + (string-split + (field ("GROUP_CONCAT(DISTINCT GeneCategory.Name SEPARATOR ';')" + GeneCategory)) + #\;)))) (define-transformer ncbi-genewiki-entries (tables (GeneRIF_BASIC - (left-join Species "USING (SpeciesId)")) - "WHERE GeneRIF_BASIC.comment IS NOT NULL AND TRIM(GeneRIF_BASIC.comment) != '' AND TRIM(GeneRIF_BASIC.symbol) != '' GROUP BY GeneRIF_BASIC.comment, GeneRIF_BASIC.createtime, GeneRIF_BASIC.VersionId, GeneRIF_BASIC.SpeciesId, GeneRIF_BASIC.TaxID") + (left-join Species "USING (SpeciesId)"))) (schema-triples (gnc:NCBIWikiEntry rdfs:subClassOf gnc:GeneWikiEntry) - (gnc:NCBIWikiEntry rdfs:comment "Represents GeneRIF Entries obtained from NCBI") - (gnt:hasVersionId a owl:ObjectProperty) - (gnt:hasVersionId rdfs:domain gnc:NCBIWikiEntry) - (gnt:hasVersionId skos:definition "The VersionId of this this resource")) + (gnc:NCBIWikiEntry rdfs:comment "Represents GeneRIF Entries obtained from NCBI")) (triples - (string->identifier - "symbol" - (regexp-substitute/global #f "[^A-Za-z0-9:]" - (field GeneRIF_BASIC symbol GeneRIFSymbol) - 'pre "_" 'post) - #:proc (lambda (x) x)) - (set rdfs:comment - (let ([ncbi-comment (sanitize-rdf-string (field GeneRIF_BASIC comment))] - [species-name - (string->identifier - "" - (remap-species-identifiers (field Species Fullname SpeciesFullName)) - #:separator "" - #:proc string-capitalize-first)] - [taxonomic-id (field GeneRIF_BASIC TaxID TaxonomicId)] - [create-time (field GeneRIF_BASIC createtime EntryCreateTime)] - [pmid (field GeneRIF_BASIC PubMed_ID PMID)] - [gene-id (field GeneRIF_BASIC GeneId)] - [version-id (field GeneRIF_BASIC VersionId)]) + (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)) + (set rdf:type + (let* ((comment (format #f "'~a'@en" + (replace-substrings + (sanitize-rdf-string + (field GeneRIF_BASIC comment)) + '(("\\" . "\\\\") + ("\n" . "\\n") + ("\r" . "\\r") + ("'" . "\\'"))))) + (create-time (format #f "~s^^xsd:datetime" + (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)) + (gene-id (field GeneRIF_BASIC GeneId)) + (taxon-id (field GeneRIF_BASIC TaxID TaxonomicId)) + (pmid (field GeneRIF_BASIC PubMed_ID)) + (version-id (field GeneRIF_BASIC versionId))) (string->symbol (string-append - "[ " - (format #f "rdf:type gnc:NCBIWikiEntry ; ") - (format #f "rdfs:comment ~s^^xsd:string ; " - ncbi-comment) - (format #f "gnt:belongsToSpecies ~a ; " - species-name) - (if (eq? #f taxonomic-id) - "" - (format #f "skos:notation taxon:~a ; " - taxonomic-id)) - (format #f "gnt:hasGeneId generif:~a ; " - gene-id) - (format #f "gnt:hasVersionId '~a'^^xsd:integer ; " - version-id) - (if (and (string? pmid) (not (string-null? pmid))) - (format #f - "~{dct:references pubmed:~a ; ~}" - (string-split pmid #\space)) - "") - (if (string? create-time) - "" - (format #f "dct:created ~s^^xsd:datetime ; " - (time-unix->string - create-time "~5"))) - " ]")))))) + (format #f "gnc:NCBIWikiEntry ;\n") + (format #f "\trdfs:label ~a ;\n" comment) + (format #f "\tgnt:belongsToSpecies ~a ;\n" species) + (format #f "\tgnt:symbol ~s ;\n" symbol) + (format #f "\tgnt:hasGeneId generif:~a ;\n" gene-id) + (match taxon-id + ((? number? x) + (format #f "\tskos:notation taxon:~a ;\n" taxon-id)) + (else "")) + (format #f "\tdct:hasVersion \"~a\"^^xsd:integer ;\n" version-id) + (format #f "\tdct:references pubmed:~a ;\n" pmid) + (format #f "\tdct:created ~a" create-time))))))) @@ -218,7 +172,6 @@ ("owl:" "<http://www.w3.org/2002/07/owl#>"))) (inputs (list - genewiki-symbols gn-genewiki-entries ncbi-genewiki-entries)) (outputs diff --git a/examples/probeset.scm b/examples/probeset.scm deleted file mode 100755 index 9f694af..0000000 --- a/examples/probeset.scm +++ /dev/null @@ -1,203 +0,0 @@ -#! /usr/bin/env guile -!# - -(use-modules (srfi srfi-1) - (srfi srfi-26) - (ice-9 format) - (ice-9 getopt-long) - (ice-9 match) - (ice-9 regex) - (transform strings) - (transform sql) - (transform triples) - (transform special-forms) - (web uri)) - - -(define-transformer probeset - (tables (ProbeSet - (left-join GeneChip "ON GeneChip.Id = ProbeSet.ChipId") - (left-join Species "ON GeneChip.SpeciesId = Species.Id")) - "WHERE ProbeSet.Name IS NOT NULL") - (schema-triples - (gnc:omimLink rdfs:Class gnc:ResourceLink) - (gnc:omimLink rdfs:label "OMIM") - (gnc:omimLink rdfs:comments "Summary from On Mendelion Inheritance in Man") - (gnc:homologeneLink rdfs:Class gnc:ResourceLink) - (gnc:homologeneLink rdfs:label "HomoloGene") - (gnc:homologeneLink rdfs:comments "Find similar genes in other species") - (gnc:uniprot a owl:ObjectProperty) - (gnc:uniprot rdfs:label "UniProt") - (gnc:uniprot rdfs:comments "UniProt resource") - (gnt:hasChip a owl:ObjectProperty) - (gnt:hasChip rdfs:domain gnc:Probeset) - (gnt:hasTargetId a owl:ObjectProperty) - (gnt:hasTargetId rdfs:domain gnc:Probeset) - (gnt:geneSymbol rdfs:domain gnc:Probeset) - (gnt:location rdfs:domain gnc:ProbeSet) - (gnt:location a owl:ObjectProperty) - (gnt:strandPosition rdfs:domain gnc:ProbeSet) - (gnt:strandPosition a owl:ObjectProperty) - (gnt:targetsRegion a owl:ObjectProperty) - (gnt:targetsRegion rdfs:domain gnc:Probeset) - (gnt:chr rdfs:domain gnc:Probeset) - (gnt:mb 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:hasBlatMbEnd a owl:ObjectProperty) - (gnt:hasBlatMbEnd 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)) - (triples - (let ((id (field ("IF(NULLIF(TRIM(ProbeSet.Name), '') IS NULL, '', TRIM(ProbeSet.Name))" - ProbeSetIdName))) - (probeset-id (field ProbeSet Id))) - (string->identifier - "probeset" - (if (string-null? id) - (number->string probeset-id) - (regexp-substitute/global - #f "[^A-Za-z0-9:]" - id - 'pre "_" 'post)))) - (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:geneSymbol - (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:location - (let* ((mb (field ProbeSet Mb)) - (chr (field ProbeSet Chr)) - (strand-probe (field ProbeSet Strand_Probe)) - (location (list chr mb))) - (match location - (("Un" mb) - (format #f "Not available")) - ((chr "") - (if (string-blank? chr) - (format #f "Not available") - (format #f "Chr ~a @ Unknown position ~a~:[~;~a~]" - chr mb - (and (string? strand-probe) (or (string=? "+" strand-probe) - (string=? "-" strand-probe))) - (cond ((string=? "+" strand-probe) - "on the plus strand") - ((string=? "-" strand-probe) - "on the minus strand") - (else ""))))) - (_ - (format #f "Chr ~a @ ~a Mb ~:[~;~a~]" - chr mb - (and (string? strand-probe) (or (string=? "+" strand-probe) - (string=? "-" strand-probe))) - (cond ((string=? "+" strand-probe) - "on the plus strand") - ((string=? "-" strand-probe) - "on the minus strand") - (else ""))))))) - (set gnt:hasGeneId - (ontology 'gene: - (string-trim-both (field ProbeSet GeneId)))) - ;; OMIM Link - (set dct:references - (let ((omim (field ProbeSet OMIM))) - (if (not (string-blank? omim)) - (string->symbol - (format #f - "<~0@*~a~1@*~a> .~%<~0@*~a~1@*~a> ~2@*~a" - "http://www.ncbi.nlm.nih.gov/omim/" - (uri-encode omim) - "a gnc:omimLink")) - ""))) - ;; Homologene Link - (set dct:references - (let ((homologene (field ProbeSet HomoloGeneID))) - (if (not (string-blank? homologene)) - (string->symbol - (format #f - "<~0@*~a~1@*~a> .~%<~0@*~a~1@*~a> ~2@*~a" - "http://www.ncbi.nlm.nih.gov/homologene/?term=" - (uri-encode homologene) - "a gnc:homologeneLink")) - ""))) - (set gnt:uniprot - (ontology 'uniprot: (field ProbeSet UniProtID))) - (set gnt:strandProbe - (field ProbeSet Strand_Probe)) - (set gnt:hasSpecificity - (field ("IFNULL(ProbeSet.Probe_set_specificity, '')" - Probe_set_specificity))) - (set gnt:hasBlatScore - (field ("IFNULL(ProbeSet.Probe_set_BLAT_score, '')" - Probe_set_BLAT_score))) - (set gnt:hasBlatMbStart - (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_start, '')" - Probe_set_Blat_Mb_start)) - '^^xsd:double)) - (set gnt:hasBlatMbEnd - (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_end, '')" - Probe_set_Blat_Mb_end)) - '^^xsd:double)) - (set gnt:hasBlatSeq (sanitize-rdf-string (field ProbeSet BlatSeq))) - (set gnt:hasTargetSeq (sanitize-rdf-string (field ProbeSet TargetSeq))))) - - - - -(let* ((option-spec - '((settings (single-char #\s) (value #t)) - (output (single-char #\o) (value #t)) - (documentation (single-char #\d) (value #t)))) - (options (getopt-long (command-line) option-spec)) - (settings (option-ref options 'settings #f)) - (output (option-ref options 'output #f)) - (documentation (option-ref options 'documentation #f)) - (%connection-settings - (call-with-input-file settings - read))) - (with-documentation - (name "ProbeSet Metadata") - (connection %connection-settings) - (table-metadata? #f) - (prefixes - '(("gn:" "<http://genenetwork.org/id/>") - ("probeset:" "<http://genenetwork.org/probeset/>") - ("gnc:" "<http://genenetwork.org/category/>") - ("gene:" "<http://www.ncbi.nlm.nih.gov/gene?cmd=Retrieve&dopt=Graphics&list_uids=>") - ("gnt:" "<http://genenetwork.org/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/>") - ("owl:" "<http://www.w3.org/2002/07/owl#>") - ("xsd:" "<http://www.w3.org/2001/XMLSchema#>") - ("qb:" "<http://purl.org/linked-data/cube#>") - ("sdmx-measure:" "<http://purl.org/linked-data/sdmx/2009/measure#>") - ("skos:" "<http://www.w3.org/2004/02/skos/core#>"))) - (inputs - (list probeset)) - (outputs - `(#:documentation ,documentation - #:rdf ,output)))) diff --git a/examples/schema.scm b/examples/schema.scm new file mode 100755 index 0000000..50cfd6a --- /dev/null +++ b/examples/schema.scm @@ -0,0 +1,70 @@ +#! /usr/bin/env guile +!# + +(use-modules (ice-9 getopt-long) + (transform triples) + (transform schema) + (transform special-forms) + (transform sql) + (transform table)) + +(define (call-with-genenetwork-database connection-settings proc) + (call-with-database "mysql" (string-join + (list (assq-ref connection-settings 'sql-username) + (assq-ref connection-settings 'sql-password) + (assq-ref connection-settings 'sql-database) + "tcp" + (assq-ref connection-settings 'sql-host) + (number->string + (assq-ref connection-settings 'sql-port))) + ":") + proc)) + +(define (transform-table-schema connection-settings db) + (let ((tables (tables connection-settings db))) + (for-each (lambda (table) + (let ((table-id (string->identifier + "table" + ;; We downcase table names in + ;; identifiers. So, we distinguish + ;; between the user and User tables. + (if (string=? (table-name table) "User") + "user2" + (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)))) + (for-each (lambda (column) + (let ((column-id (column-id (table-name table) + (column-name column)))) + (triple column-id 'rdf:type 'gn:sqlTableField) + (triple column-id 'gn:name (column-name column)) + (triple column-id 'gn:sqlFieldType (column-type column)) + (triple table-id 'gn:hasField column-id))) + (table-columns table)))) + tables))) + + +(let* ((option-spec + '((settings (single-char #\s) (value #t)) + (output (single-char #\o) (value #t)) + (documentation (single-char #\d) (value #t)))) + (options (getopt-long (command-line) option-spec)) + (settings (option-ref options 'settings #f)) + (output (option-ref options 'output #f)) + (documentation (option-ref options 'documentation #f)) + (%connection-settings (call-with-input-file settings read))) + (call-with-genenetwork-database + %connection-settings + (lambda (db) + (with-output-to-file output + (lambda () + (prefix "rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>") + (prefix "rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>") + (prefix "gn:" "<http://genenetwork.org/id/>") + (prefix "gnc:" "<http://genenetwork.org/category/>") + (prefix "gnt:" "<http://genenetwork.org/term/>") + (prefix "xsd:" "<http://www.w3.org/2001/XMLSchema#>") + (prefix "owl:" "<http://www.w3.org/2002/07/owl#>") + (newline) + (transform-table-schema %connection-settings db)))))) |