about summary refs log tree commit diff
diff options
context:
space:
mode:
Diffstat
-rwxr-xr-xexamples/datasets.scm2
-rwxr-xr-xexamples/genotype-datasets.scm53
-rwxr-xr-xexamples/genotype.scm70
-rwxr-xr-xexamples/ontology.scm39
-rwxr-xr-xexamples/phenotype-datasets.scm54
-rwxr-xr-xexamples/phenotype.scm53
-rwxr-xr-xload-rdf.scm2
-rw-r--r--transform/strings.scm4
8 files changed, 200 insertions, 77 deletions
diff --git a/examples/datasets.scm b/examples/datasets.scm
index 55d8c35..e4f35da 100755
--- a/examples/datasets.scm
+++ b/examples/datasets.scm
@@ -25,6 +25,7 @@
     (set rdf:type 'dcat:Dataset)
     (set dct:title (normalize-string-field (field InfoFiles InfoPageName)))
     (set dct:identifier (format #f "GN~a" (field InfoFiles GN_AccesionId)))
+    (set gnt:has_genotype_files (string->symbol (format #f "gn-files:GN~a" (field InfoFiles GN_AccesionId))))
     (set gnt:has_strain
          (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))
     (set gnt:has_experiment_type
@@ -105,6 +106,7 @@
       ("gn:" "<http://rdf.genenetwork.org/v1/id/>")
       ("gnc:" "<http://rdf.genenetwork.org/v1/category/>")
       ("gnt:" "<http://rdf.genenetwork.org/v1/term/>")
+      ("gn-files:" "<http://files.genenetwork.org/current/>")
       ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>")
       ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>")
       ("owl:" "<http://www.w3.org/2002/07/owl#>")
diff --git a/examples/genotype-datasets.scm b/examples/genotype-datasets.scm
index f140600..ebe2349 100755
--- a/examples/genotype-datasets.scm
+++ b/examples/genotype-datasets.scm
@@ -15,29 +15,39 @@
 
 
 (define-transformer gn:set->gn:dataset
-  (tables (Datasets
-           (inner-join InfoFiles "ON InfoFiles.DatasetId = Datasets.DatasetId")
-           (inner-join InbredSet "ON InbredSet.Id = InfoFiles.InbredSetId")
+  (tables (Species
+           (inner-join InbredSet "ON InbredSet.SpeciesId = Species.Id")
            (inner-join GenoFreeze "ON GenoFreeze.InbredSetId = InbredSet.Id"))
-          "WHERE GenoFreeze.public > 0 GROUP BY Datasets.DatasetId")
-  (schema-triples
-   (gnt:has_genotype_data rdf:type owl:ObjectProperty)
-   (gnt:has_genotype_data rdfs:label "this resources has genotype data.")
-   (gnt:has_genotype_data rdfs:comment "Associates a resource with its genotype data.")
-   (gnt:has_genotype_data rdfs:domain gnc:set)
-   (gnt:has_genotype_data rdfs:range dcat:Dataset)
-   (gnt:has_genotype_data rdfs:subPropertyOf dct:relation))
+          "WHERE GenoFreeze.public > 0 AND Species.Name != 'monkey' GROUP BY Species.Name, GenoFreeze.ShortName")
   (triples (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_")
-    (set gnt:has_genotype_data (string->identifier "dataset" (field GenoFreeze Name) #:separator "_"))))
+    (multiset gnt:has_genotype_data
+              (map (cut string->identifier "dataset" <> #:separator "_")
+                   (string-split
+                    (field ("GROUP_CONCAT(GenoFreeze.Name SEPARATOR ',')"
+                            dataset_name))
+                    #\,)))))
 
-(define-transformer gn:dataset->gn:set
-  (tables (Datasets
-           (inner-join InfoFiles "ON InfoFiles.DatasetId = Datasets.DatasetId")
-           (inner-join InbredSet "ON InbredSet.Id = InfoFiles.InbredSetId")
-           (inner-join GenoFreeze "ON GenoFreeze.InbredSetId = InbredSet.Id"))
-          "WHERE GenoFreeze.public > 0 GROUP BY Datasets.DatasetId")
+(define-transformer gn:dataset->metadata
+  (tables (GenoFreeze
+           (inner-join InbredSet "ON InbredSet.Id = GenoFreeze.InbredSetId")
+           (inner-join Species "ON InbredSet.SpeciesId = Species.Id"))
+          "WHERE GenoFreeze.public > 0 AND Species.Name != 'monkey'")
+  (triples (string->identifier "dataset" (field GenoFreeze Name) #:separator "_")
+    (set gnt:has_strain (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))
+    (set dct:created (annotate-field (field GenoFreeze CreateTime) '^^xsd:datetime))))
+
+(define-transformer gn:dataset->marker/snp-count
+  (tables (GenoFreeze
+           (inner-join InbredSet "ON InbredSet.Id = GenoFreeze.InbredSetId")
+           (inner-join Species "ON InbredSet.SpeciesId = Species.Id")
+           (inner-join Geno "ON Geno.SpeciesId = Species.Id"))
+          "WHERE GenoFreeze.public > 0 AND Species.Name != 'monkey' GROUP BY GenoFreeze.Name")
   (triples (string->identifier "dataset" (field GenoFreeze Name) #:separator "_")
-    (set gnt:has_strain (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))))
+    (set gnt:has_marker_count
+         (string->symbol
+          (format #f "'~s'^^xsd:integer"
+                  (field
+                   ("COUNT(DISTINCT Geno.Marker_Name)" MarkerCount)))))))
 
 
 (let* ((option-spec
@@ -68,7 +78,10 @@
       ("xkos:" "<http://rdf-vocabulary.ddialliance.org/xkos#>")
       ("xsd:" "<http://www.w3.org/2001/XMLSchema#>")))
    (inputs
-    (list gn:set->gn:dataset gn:dataset->gn:set))
+    (list
+     gn:set->gn:dataset
+     gn:dataset->metadata
+     gn:dataset->marker/snp-count))
    (outputs
     `(#:documentation ,documentation
       #:rdf ,output))))
diff --git a/examples/genotype.scm b/examples/genotype.scm
new file mode 100755
index 0000000..4a91b63
--- /dev/null
+++ b/examples/genotype.scm
@@ -0,0 +1,70 @@
+#! /usr/bin/env guile
+!#
+
+(use-modules (rnrs programs)
+             (rnrs io ports)
+             (srfi srfi-1)
+             (srfi srfi-26)
+             (ice-9 getopt-long)
+             (ice-9 match)
+             (ice-9 regex)
+             (transform strings)
+             (transform sql)
+             (transform triples)
+             (transform special-forms))
+
+(define-transformer gn:markers/snps->metadata
+  (tables (Geno
+           (inner-join Species "ON Geno.SpeciesId = Species.Id"))
+          "WHERE Species.Name != 'monkey'")
+  (triples (string->identifier "marker" (field Geno Name) #:separator "_")
+    (set gnt:has_species
+         (string->identifier "" (remap-species-identifiers (field Species Fullname))))
+    (set rdf:type 'gnc:dna_marker)
+    (set skos:prefLabel (field Geno Name))
+    (set skos:altLabel (field Geno Marker_Name))
+    (set gnt:chr (field Geno Chr))
+    (set gnt:mb (annotate-field (field Geno Mb) '^^xsd:doubleg))
+    (set gnt:sequence (field Geno Sequence))
+    (set gnt:source (field Geno Source))
+    (set rdfs:comment (field Geno Comments))))
+
+
+
+(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 "Phenotypes Metadata")
+   (connection %connection-settings)
+   (table-metadata? #f)
+   (prefixes
+    '(("dcat:" "<http://www.w3.org/ns/dcat#>")
+      ("dct:" "<http://purl.org/dc/terms/>")
+      ("gn:" "<http://rdf.genenetwork.org/v1/id/>")
+      ("owl:" "<http://www.w3.org/2002/07/owl#>")
+      ("gnc:" "<http://rdf.genenetwork.org/v1/category/>")
+      ("gnd:" "<https://cd.genenetwork.org/lmdb/v1/data/traits/>")
+      ("gnt:" "<http://rdf.genenetwork.org/v1/term/>")
+      ("sdmx-measure:" "<http://purl.org/linked-data/sdmx/2009/measure#>")
+      ("skos:" "<http://www.w3.org/2004/02/skos/core#>")
+      ("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#>")
+      ("qb:" "<http://purl.org/linked-data/cube#>")
+      ("xkos:" "<http://rdf-vocabulary.ddialliance.org/xkos#>")
+      ("pubmed:" "<http://rdf.ncbi.nlm.nih.gov/pubmed/>")))
+   (inputs
+    (list gn:markers/snps->metadata))
+   (outputs
+    `(#:documentation ,documentation
+      #:rdf ,output))))
+
diff --git a/examples/ontology.scm b/examples/ontology.scm
index 2e03c53..724a75a 100755
--- a/examples/ontology.scm
+++ b/examples/ontology.scm
@@ -38,7 +38,7 @@
       (triple 'gnc:population_category 'a 'xkos:ClassificationLevel)
       (triple 'gnc:population_category 'rdfs:label "Population Category")
       (triple 'gnc:population_category 'skos:inScheme 'gnc:resource_classification_scheme)
-      (triple 'gnc:population_category 'skos:prefLabel "Species")
+      (triple 'gnc:population_category 'skos:prefLabel "Population Category")
       (triple 'gnc:population_category 'xkos:depth "3")
       (triple 'gnc:population_category 'xkos:nextLevel 'gnc:set)
       (triple 'gnc:population_category 'xkos:previousLevel 'gnc:species)
@@ -234,4 +234,39 @@
       (triple 'gnt:submitter 'skos:definition "A person who submitted this resource to GN")
       (triple 'gnt:has_phenotype_data 'a 'owl:ObjectProperty)
       (triple 'gnt:has_phenotype_data 'rdfs:domain 'gnc:set)
-      (triple 'gnt:has_phenotype_data 'skos:definition "This resource has phenotype data."))))
+      (triple 'gnt:has_phenotype_data 'skos:definition "This resource has phenotype data.")
+
+      ;; Genotypes
+      (triple 'gnc:dna_marker 'a 'owl:Class)
+      (triple 'gnc:dna_marker 'a 'skos:Concept)
+      (triple 'gnc:dna_marker 'rdfs:label "A DNA Marker or SNP")
+      (triple 'gnt:has_genotype_files 'rdfs:label "This resource has these genotype files")
+      (triple 'gnt:has_genotype_files 'rdfs:domain 'dcat:Dataset)
+      (triple 'gnt:has_genotype_data 'rdf:type 'owl:ObjectProperty)
+      (triple 'gnt:has_genotype_data 'rdfs:label "this resources has genotype data.")
+      (triple 'gnt:has_genotype_data 'rdfs:comment "Associates a resource with its genotype data.")
+      (triple 'gnt:has_genotype_data 'rdfs:domain 'gnc:set)
+      (triple 'gnt:has_genotype_data 'rdfs:range 'dcat:Dataset)
+      (triple 'gnt:has_genotype_data 'rdfs:subPropertyOf 'dct:relation)
+      (triple 'gnt:has_marker_count 'rdf:type 'owl:ObjectProperty)
+      (triple 'gnt:has_marker_count 'rdfs:label "this resources has N number of dna markers/SNPs.")
+      (triple 'gnt:has_marker_count 'rdfs:domain 'xsd:integer)
+      (triple 'gnt:has_marker_count 'rdfs:range 'dcat:Dataset)
+      (triple 'gnt:chr 'a 'qb:MeasureProperty)
+      (triple 'gnt:chr 'a 'rdf:Property)
+      (triple 'gnt:chr 'rdfs:label "Chromosome")
+      (triple 'gnt:chr 'rdfs:domain 'gnc:marker)
+      (triple 'gnt:chr 'rdfs:range 'rdfs:Literal)
+      (triple 'gnt:chr 'rdfs:subPropertyOf 'sdmx-measure:obsValue)
+      (triple 'gnt:mb 'rdfs:label "Megabase")
+      (triple 'gnt:mb 'rdfs:domain 'gnc:marker)
+      (triple 'gnt:mb 'rdfs:range 'rdfs:Literal)
+      (triple 'gnt:mb 'rdfs:subPropertyOf 'sdmx-measure:obsValue)
+      (triple 'gnt:sequence 'rdfs:label "Sequence")
+      (triple 'gnt:sequence 'rdfs:domain 'gnc:marker)
+      (triple 'gnt:sequence 'rdfs:range 'rdfs:Literal)
+      (triple 'gnt:sequence 'rdfs:subPropertyOf 'sdmx-measure:obsValue)
+      (triple 'gnt:source 'rdfs:label "Source")
+      (triple 'gnt:source 'rdfs:domain 'gnc:marker)
+      (triple 'gnt:source 'rdfs:range 'rdfs:Literal)
+      (triple 'gnt:source 'rdfs:subPropertyOf 'sdmx-measure:obsValue))))
diff --git a/examples/phenotype-datasets.scm b/examples/phenotype-datasets.scm
index 410178a..4819627 100755
--- a/examples/phenotype-datasets.scm
+++ b/examples/phenotype-datasets.scm
@@ -15,13 +15,17 @@
 
 
 (define-transformer gn:set->gn:dataset
-  (tables (Datasets
-           (inner-join InfoFiles "ON InfoFiles.DatasetId = Datasets.DatasetId")
-           (inner-join InbredSet "ON InbredSet.Id = InfoFiles.InbredSetId")
+  (tables (Species
+           (inner-join InbredSet "ON InbredSet.SpeciesId = Species.Id")
            (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id"))
-          "WHERE PublishFreeze.public > 0 GROUP BY Datasets.DatasetId")
+          "WHERE PublishFreeze.public > 0 AND Species.Name != 'monkey' GROUP BY Species.Name, PublishFreeze.ShortName")
   (triples (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_")
-    (set gnt:has_phenotype_data (string->identifier "dataset" (field PublishFreeze Name) #:separator "_"))))
+    (multiset gnt:has_phenotype_data
+              (map (cut string->identifier "dataset" <> #:separator "_")
+                   (string-split
+                    (field ("GROUP_CONCAT(PublishFreeze.Name SEPARATOR ',')"
+                            dataset_name))
+                    #\,)))))
 
 (define-transformer gn:dataset->gn:set
   (tables (Datasets
@@ -32,6 +36,40 @@
   (triples (string->identifier "dataset" (field PublishFreeze Name) #:separator "_")
     (set gnt:has_strain (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))))
 
+(define-transformer gn:dataset->metadata
+  (tables (PublishXRef
+           (inner-join InbredSet "ON InbredSet.InbredSetId = PublishXRef.InbredSetId")
+           (inner-join Species "ON InbredSet.SpeciesId = Species.Id")
+           (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id")
+           (inner-join Publication "ON Publication.Id = PublishXRef.PublicationId")
+           (inner-join Phenotype "ON Phenotype.Id = PublishXRef.PhenotypeId"))
+          "WHERE InbredSet.public > 0 GROUP BY Species.Name, PublishFreeze.Name")
+  (triples (string->identifier "dataset" (field PublishFreeze Name) #:separator "_")
+    (set dct:created (annotate-field (field PublishFreeze CreateTime) '^^xsd:datetime))
+    (set gnt:has_strain (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))))
+
+(define-transformer gn:dataset->gn:trait
+  (tables (PublishXRef
+           (inner-join InbredSet "ON InbredSet.InbredSetId = PublishXRef.InbredSetId")
+           (inner-join Species "ON InbredSet.SpeciesId = Species.Id")
+           (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id")
+           (inner-join Publication "ON Publication.Id = PublishXRef.PublicationId")
+           (inner-join Phenotype "ON Phenotype.Id = PublishXRef.PhenotypeId"))
+          "WHERE InbredSet.public > 0")
+  (triples (string->identifier "dataset" (field PublishFreeze Name) #:separator "_")
+    (set gnt:has_phenotype_trait
+         (let ((post-abbrev (blank-p (field Phenotype Post_publication_abbreviation)))
+               (pre-abbrev (blank-p (field Phenotype Pre_publication_abbreviation)))
+               (post-desc (blank-p (field Phenotype Post_publication_description)))
+               (pre-desc (blank-p (field Phenotype Post_publication_description))))
+           (string->identifier
+            "trait"
+            (format #f "~a_~a" (field PublishFreeze Name)
+                    (or post-abbrev pre-abbrev post-desc pre-desc))
+            #:separator "_")))
+    (set dct:created (annotate-field (field PublishFreeze CreateTime) '^^xsd:datetime))
+    (set gnt:has_strain (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))))
+
 
 (let* ((option-spec
         '((settings (single-char #\s) (value #t))
@@ -61,7 +99,11 @@
       ("xkos:" "<http://rdf-vocabulary.ddialliance.org/xkos#>")
       ("xsd:" "<http://www.w3.org/2001/XMLSchema#>")))
    (inputs
-    (list gn:set->gn:dataset gn:dataset->gn:set))
+    (list
+     gn:set->gn:dataset
+     gn:dataset->gn:set
+     gn:dataset->metadata
+     gn:dataset->gn:trait))
    (outputs
     `(#:documentation ,documentation
       #:rdf ,output))))
diff --git a/examples/phenotype.scm b/examples/phenotype.scm
index 72d3ed3..834ff5a 100755
--- a/examples/phenotype.scm
+++ b/examples/phenotype.scm
@@ -12,57 +12,14 @@
              (transform sql)
              (transform triples)
              (transform special-forms))
+
 
-(define (blank-p str)
-  (if (string-blank? str) #f str))
+
 
 
-(define-transformer gn:set->gn:dataset
-  (tables (Species
-           (inner-join InbredSet "ON InbredSet.SpeciesId = Species.Id")
-           (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id"))
-          "WHERE PublishFreeze.public > 0 AND Species.Name != 'monkey' GROUP BY Species.Name, PublishFreeze.ShortName")
-  (triples (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_")
-    (multiset gnt:has_phenotype_data
-              (map (cut string->identifier "dataset" <> #:separator "_")
-                   (string-split
-                    (field ("GROUP_CONCAT(PublishFreeze.Name SEPARATOR ',')"
-                            dataset_name))
-                    #\,)))))
 
-(define-transformer gn:dataset->metadata
-  (tables (PublishXRef
-           (inner-join InbredSet "ON InbredSet.InbredSetId = PublishXRef.InbredSetId")
-           (inner-join Species "ON InbredSet.SpeciesId = Species.Id")
-           (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id")
-           (inner-join Publication "ON Publication.Id = PublishXRef.PublicationId")
-           (inner-join Phenotype "ON Phenotype.Id = PublishXRef.PhenotypeId"))
-          "WHERE InbredSet.public > 0 GROUP BY Species.Name, PublishFreeze.Name")
-  (triples (string->identifier "dataset" (field PublishFreeze Name) #:separator "_")
-    (set dct:created (annotate-field (field PublishFreeze CreateTime) '^^xsd:datetime))
-    (set gnt:has_strain (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))))
 
-(define-transformer gn:dataset->gn:trait
-  (tables (PublishXRef
-           (inner-join InbredSet "ON InbredSet.InbredSetId = PublishXRef.InbredSetId")
-           (inner-join Species "ON InbredSet.SpeciesId = Species.Id")
-           (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id")
-           (inner-join Publication "ON Publication.Id = PublishXRef.PublicationId")
-           (inner-join Phenotype "ON Phenotype.Id = PublishXRef.PhenotypeId"))
-          "WHERE InbredSet.public > 0")
-  (triples (string->identifier "dataset" (field PublishFreeze Name) #:separator "_")
-    (set gnt:has_phenotype_trait
-         (let ((post-abbrev (blank-p (field Phenotype Post_publication_abbreviation)))
-               (pre-abbrev (blank-p (field Phenotype Pre_publication_abbreviation)))
-               (post-desc (blank-p (field Phenotype Post_publication_description)))
-               (pre-desc (blank-p (field Phenotype Post_publication_description))))
-           (string->identifier
-            "trait"
-            (format #f "~a_~a" (field PublishFreeze Name)
-                    (or post-abbrev pre-abbrev post-desc pre-desc))
-            #:separator "_")))
-    (set dct:created (annotate-field (field PublishFreeze CreateTime) '^^xsd:datetime))
-    (set gnt:has_strain (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))))
+
 
 (define-transformer gnc:phenotype->gn:phenotype
   (tables (Phenotype))
@@ -198,9 +155,7 @@
       ("xkos:" "<http://rdf-vocabulary.ddialliance.org/xkos#>")
       ("pubmed:" "<http://rdf.ncbi.nlm.nih.gov/pubmed/>")))
    (inputs
-    (list gn:dataset->metadata
-          gn:dataset->gn:trait
-          gnc:phenotype->gn:phenotype
+    (list gnc:phenotype->gn:phenotype
           gn:phenotype->metadata
           gn:trait->gn:phenotype))
    (outputs
diff --git a/load-rdf.scm b/load-rdf.scm
index 590fbc0..44bda35 100755
--- a/load-rdf.scm
+++ b/load-rdf.scm
@@ -111,6 +111,8 @@ DB.DBA.XML_SET_NS_DECL ('v', 'http://www.w3.org/2006/vcard/ns#', 2);
 DB.DBA.XML_SET_NS_DECL ('xkos', 'http://rdf-vocabulary.ddialliance.org/xkos#', 2);
 DB.DBA.XML_SET_NS_DECL ('schema', 'https://schema.org/', 2);
 DB.DBA.XML_SET_NS_DECL ('foaf', 'http://xmlns.com/foaf/0.1/#term_', 2);
+DB.DBA.XML_SET_NS_DECL ('gnd', 'https://cd.genenetwork.org/lmdb/v1/data/traits/', 2);
+DB.DBA.XML_SET_NS_DECL ('gn-files:', 'http://files.genenetwork.org/current/', 2);
 "
              port
              password))
diff --git a/transform/strings.scm b/transform/strings.scm
index 51c5ed1..7b62349 100644
--- a/transform/strings.scm
+++ b/transform/strings.scm
@@ -17,8 +17,12 @@
             string-capitalize-first
             normalize-string-field
             fix-email-id
+            blank-p
             investigator-attributes->id))
 
+(define (blank-p str)
+  (if (string-blank? str) #f str))
+
 (define (lower-case-and-replace-spaces str)
   (string-map
    (lambda (c)
generated by cgit-pink 1.4.1 (git 2.36.1) at 2026-02-06 06:38:29 +0000