Touches #648: add cnv classes to graphql. Some minor formatting and o…

…ther changes.

resources/class-names.edn

@@ -102,7 +102,7 @@
  [:ga4gh/VariationDescriptor "https://terms.ga4gh.org/VariationDescriptor"]
  [:ga4gh/CanonicalReference "https://terms.ga4gh.org/CanonicalReference"]
  [:geno/SequenceInterval "http://purl.obolibrary.org/obo/GENO_0000965"]
- [:geno/SequenceFeatureLocation"http://purl.obolibrary.org/obo/GENO_0000815"]
+ [:geno/SequenceFeatureLocation "http://purl.obolibrary.org/obo/GENO_0000815"]
  [:geno/SequenceFeatureSet "http://purl.obolibrary.org/obo/GENO_0000659"]
  [:geno/PathogenicForCondition "http://purl.obolibrary.org/obo/GENO_0000840"]
  [:geno/BenignForCondition "http://purl.obolibrary.org/obo/GENO_0000843"]
@@ -152,7 +152,7 @@
  [:sepio/ClinGenGeneValidityEvaluationCriteriaSOP7 "http://purl.obolibrary.org/obo/SEPIO_0004095"]
  [:sepio/ClinGenGeneValidityEvaluationCriteriaSOP8 "http://purl.obolibrary.org/obo/SEPIO_0004096"]
  [:sepio/ClinGenGeneValidityEvaluationCriteriaSOP9 "http://purl.obolibrary.org/obo/SEPIO_0004171"]
-[:sepio/ClinGenProteinFunctionCriteria
+ [:sepio/ClinGenProteinFunctionCriteria
   "http://purl.obolibrary.org/obo/SEPIO_0004074"]
  [:sepio/NonHumanModelOrganismEvidenceItem
   "http://purl.obolibrary.org/obo/SEPIO_0004046"]
@@ -341,9 +341,9 @@
  [:sepio/ProbandEvidenceLine "http://purl.obolibrary.org/obo/SEPIO_0004097"]
  [:sepio/ProbandScoreCapEvidenceLine "http://purl.obolibrary.org/obo/SEPIO_0004174"]
  [:sepio/ProteinInteractionGeneticInteractionEvidenceLine
-   "http://purl.obolibrary.org/obo/SEPIO_0004182"]
+  "http://purl.obolibrary.org/obo/SEPIO_0004182"]
  [:sepio/ProteinInteractionNegativeGeneticInteractionEvidenceLine
-   "http://purl.obolibrary.org/obo/SEPIO_0004183"]
+  "http://purl.obolibrary.org/obo/SEPIO_0004183"]
  [:sepio/ProteinInteractionPhysicalAssociationEvidenceLine
   "http://purl.obolibrary.org/obo/SEPIO_0004184"]
  [:sepio/ProteinInteractionPositiveGeneticInteractionEvidenceLine
@@ -355,7 +355,7 @@
  [:sepio/NullVariantEvidenceItem "http://purl.obolibrary.org/obo/SEPIO_0004117"]
  [:sepio/NonNullVariantEvidenceItem "http://purl.obolibrary.org/obo/SEPIO_0004118"]
  [:sepio/VariantFunctionalImpactEvidenceItem "http://purl.obolibrary.org/obo/SEPIO_0004119"]
- 
+
  ;;
  [:sepio/GeneValidityEvidenceLine
   "http://purl.obolibrary.org/obo/SEPIO_0004082"]
@@ -407,6 +407,7 @@
  [:vrs/Allele "https://vrs.ga4gh.org/terms/Allele"]
  [:vrs/Text "https://vrs.ga4gh.org/terms/Text"]
  [:vrs/CanonicalVariation "https://vrs.ga4gh.org/terms/CanonicalVariation"]
+ [:vrs/AbsoluteCopyNumber "https://vrs.ga4gh.org/terms/AbsoluteCopyNumber"]
  [:vrs/TextUtilityVariation "https://vrs.ga4gh.org/terms/TextUtilityVariation"]
  [:vrs/VariationDescriptor "https://vrs.ga4gh.org/terms/VariationDescriptor"]
  [:vrs/CategoricalVariationDescriptor "https://vrs.ga4gh.org/terms/CategoricalVariationDescriptor"]
@@ -420,4 +421,3 @@
  [:vrs/SequenceInterval "https://vrs.ga4gh.org/terms/SequenceInterval"]
  [:chebi/Drug "http://purl.obolibrary.org/obo/CHEBI_23888"]
  [:stato/Cohort "http://purl.obolibrary.org/obo/STATO_0000203"]]
-

resources/property-names.edn

@@ -245,6 +245,8 @@
  [:vrs/end "https://vrs.ga4gh.org/terms/end"]
  [:vrs/sequence_id "https://vrs.ga4gh.org/terms/end"]
  [:vrs/definition "https://vrs.ga4gh.org/terms/definition"]
+ [:vrs/subject "https://vrs.ga4gh.org/terms/subject"]
+ [:vrs/copies "https://vrs.ga4gh.org/terms/copies"]
  [:cnt/chars "http://www.w3.org/2011/content#chars"]
  [:data/genome-build-identifier "http://edamontology.org/data_2340"]
  [:dc/abstract "http://purl.org/dc/terms/abstract"]
@@ -1267,7 +1269,7 @@
  [:sepio/lower-confidence-limit
   "http://purl.obolibrary.org/obo/SEPIO_0000306"]
  [:sepio/paternity-maternity-confirmed
-  "http://purl.obolibrary.org/obo/SEPIO_0004161"] 
+  "http://purl.obolibrary.org/obo/SEPIO_0004161"]
  [:sepio/measures-parameter
   "http://purl.obolibrary.org/obo/SEPIO_0000114"]
  [:sepio/meets-inclusion-criteria "http://purl.obolibrary.org/obo/SEPIO_0004126"]
@@ -1321,7 +1323,7 @@
  [:sepio/score-status "http://purl.obolibrary.org/obo/SEPIO_0004130"]
  [:sepio/statistical-significance
   "http://purl.obolibrary.org/obo/SEPIO_0004164"]
-  [:sepio/statistical-significance-type
+ [:sepio/statistical-significance-type
   "http://purl.obolibrary.org/obo/SEPIO_0004165"]
  [:sepio/statistical-significance-value-type
   "http://purl.obolibrary.org/obo/SEPIO_0004166"]

src/genegraph/source/graphql/experimental_schema.clj

@@ -42,6 +42,7 @@
    model-variation/vrs-allele
    model-variation/vrs-text
    model-variation/vrs-canonical-variation
+   model-variation/vrs-absolute-copy-number
    model-variation/vrs-literal-sequence-expression
    model-variation/vrs-sequence-location
    model-variation/vrs-sequence-interval

src/genegraph/source/graphql/schema/types.clj

@@ -14,6 +14,7 @@
     [:vrs/Allele :Allele]
     [:vrs/Text :Text]
     [:vrs/CanonicalVariation :CanonicalVariation]
+    [:vrs/AbsoluteCopyNumber :AbsoluteCopyNumber]
     [:vrs/SequenceLocation :SequenceLocation]
     [:vrs/SequenceInterval :SequenceInterval]
     [:vrs/LiteralSequenceExpression :LiteralSequenceExpression]
@@ -27,4 +28,3 @@
     [:sepio/ValueSet :ValueSet]
     [:pco/Family :Family]]
    :default-type-mapping :GenericResource})
-

src/genegraph/source/graphql/schema/variation_descriptor.clj

@@ -34,6 +34,16 @@
             :state {:type :LiteralSequenceExpression
                     :path [:vrs/state]}}})
 
+(def vrs-absolute-copy-number
+  {:name :AbsoluteCopyNumber
+   :graphql-type :object
+   :description "GA4GH Absolute Copy Number Variation"
+   :implements [:Resource]
+   :fields {:subject {:type :SequenceLocation
+                      :path [:vrs/subject]}
+            :copies {:type :Number
+                     :path [:vrs/copies]}}})
+
 (def vrs-text
   {:name :Text
    :graphql-type :object
@@ -119,7 +129,8 @@
    :graphql-type :object
    :description "Descriptor for a categorical variation"
    :implements [:Resource]
-   :fields {:value {:type :CanonicalVariation
+   :fields {:value {:type :Resource
+                    :description "CanonicalVariation or AbsoluteCopyNumber variation"
                     :path [:rdf/value]}
             :xrefs {:type '(list String)
                     :path [:vrs/xrefs]}

src/genegraph/transform/clinvar/jsonld/clinical_assertion.clj

@@ -57,8 +57,7 @@ WHERE {
     FILTER(?other_variant_release_date > ?variant_release_date)
   }
 }
-ORDER BY ?s_variant ?gene_id"
-  )
+ORDER BY ?s_variant ?gene_id")
 
 (defn get-genes-for-clinical-assertion
   "Returns the genes associated with this clinical assertion through the variation"
@@ -115,16 +114,14 @@ ORDER BY ?s_variant ?gene_id"
           (do (log/info :msg "Assertion classification context is :PHARMACOGENOMIC")
               :PHARMACOGENOMIC)
 
-          (or (in? ["practice guideline" "reviewed by expert panel"] review-status)
+          (or (in? review-status ["practice guideline" "reviewed by expert panel"])
               (let [{normalized-clinsig :normalized normalized-group :group} (normalize-clinvar-clinsig clinsig)]
                 (log/debug :msg (format "Normalized clinsig %s to %s" clinsig normalized-clinsig))
                 (= "path" normalized-group)))
           (do (log/info :msg "Assertion classification context is :GERMLINE_DISEASE")
               :GERMLINE_DISEASE)
 
-          :default :OTHER
-          ))
-  )
+          :default :OTHER)))
 
 (defn clinical-assertion-to-jsonld [msg]
   (let [id (format (str iri/clinvar-assertion "%s.%s")
@@ -138,75 +135,70 @@ ORDER BY ?s_variant ?gene_id"
                              "clinvar" "https://www.ncbi.nlm.nih.gov/clinvar/"
                              ;rdf-type          {"@type" "@id"}
                              ;:cg/ClinVarObject {"@type" "@id"}
-                             }
-                 }
-        msg (assoc msg :cg/classification-context (compute-clingen-classification-context msg))
-        ]
+                             }}
+        msg (assoc msg :cg/classification-context (compute-clingen-classification-context msg))]
     (genegraph-kw-to-iri
-      (merge
-        context
+     (merge
+      context
         ; TODO Add @base
-        {"@type" [:cg/ClinVarObject
-                  (str iri/cgterms "EvidenceLine")],
-         "@id" evidence-line-id,
+      {"@type" [:cg/ClinVarObject
+                (str iri/cgterms "EvidenceLine")],
+       "@id" evidence-line-id,
 
-         :sepio/has-evidence-direction "supports"
-         :sepio/evidence-line-strength (scv-review-status-to-evidence-strength-map
-                                         (:review_status msg))
+       :sepio/has-evidence-direction "supports"
+       :sepio/evidence-line-strength (scv-review-status-to-evidence-strength-map
+                                      (:review_status msg))
          ; The SCV itself is an evidence item within an evidence line that pertains to a variant assertion
-         :sepio/has-evidence-item
-         (merge
-           {"@type" [:cg/ClinVarObject
-                     assertion-rdf-type]
-            "@id" id
-            :dc/is-version-of {"@id" (str iri/clinvar-assertion (:id msg))}
-            :dc/has-version (:version msg)
-            :dc/title (:title msg)
+       :sepio/has-evidence-item
+       (merge
+        {"@type" [:cg/ClinVarObject
+                  assertion-rdf-type]
+         "@id" id
+         :dc/is-version-of {"@id" (str iri/clinvar-assertion (:id msg))}
+         :dc/has-version (:version msg)
+         :dc/title (:title msg)
 
-            :sepio/has-subject {"@id" (str iri/clinvar-variation (:variation_id msg))}
-            :sepio/has-predicate (:interpretation_description msg)
-            :sepio/has-object (str iri/trait-set (:trait_set_id msg))
-            :sepio/date-created (:date_created msg)
-            :sepio/date-updated (:date_last_updated msg)
+         :sepio/has-subject {"@id" (str iri/clinvar-variation (:variation_id msg))}
+         :sepio/has-predicate (:interpretation_description msg)
+         :sepio/has-object (str iri/trait-set (:trait_set_id msg))
+         :sepio/date-created (:date_created msg)
+         :sepio/date-updated (:date_last_updated msg)
 
-            :sepio/qualified-contribution {:sepio/activity-date (:interpretation_date_last_evaluated msg)
-                                           :sepio/has-role "SubmitterRole"
-                                           :sepio/has-agent {"@id" (str iri/submitter (:submitter_id msg))}}
+         :sepio/qualified-contribution {:sepio/activity-date (:interpretation_date_last_evaluated msg)
+                                        :sepio/has-role "SubmitterRole"
+                                        :sepio/has-agent {"@id" (str iri/submitter (:submitter_id msg))}}
 
             ; ClinGen/ClinVar additional renamed terms (namespaced to @vocab)
-            "allele_origin" (:allele_origins msg)
-            "collection_method" (:collection_methods msg)
-            "submitted_condition" (str iri/clinical-assertion-trait-set (:clinical_assertion_trait_set_id msg))
+         "allele_origin" (:allele_origins msg)
+         "collection_method" (:collection_methods msg)
+         "submitted_condition" (str iri/clinical-assertion-trait-set (:clinical_assertion_trait_set_id msg))
             ; TODO update field name if change occurs here https://github.com/clingen-data-model/clinvar-streams/issues/3
-            "submitted_variation" (:clinical_assertion_variations msg)
-            }
-           (-> msg (dissoc
-                     :id
-                     :version
-                     :title
-                     :variation_id
-                     :interpretation_description
-                     :trait_set_id
-                     :date_created
-                     :date_last_updated
-                     :interpretation_date_last_evaluated
-                     :submitter_id
-                     :allele_origins
-                     :collection_methods
-                     :clinical_assertion_trait_set_id
-                     :clinical_assertion_variations))),     ; End assertion
+         "submitted_variation" (:clinical_assertion_variations msg)}
+        (-> msg (dissoc
+                 :id
+                 :version
+                 :title
+                 :variation_id
+                 :interpretation_description
+                 :trait_set_id
+                 :date_created
+                 :date_last_updated
+                 :interpretation_date_last_evaluated
+                 :submitter_id
+                 :allele_origins
+                 :collection_methods
+                 :clinical_assertion_trait_set_id
+                 :clinical_assertion_variations))),     ; End assertion
 
          ; Reverse relation from evidence line to parent variation archive
-         "@reverse" {:sepio/has-evidence-line
-                     [{"@id" (let [variation-archive-iri (format (str iri/variation-archive "%s")
-                                                                 (:variation_archive_id msg))]
-                               (if (empty? (:variation_archive_id msg))
-                                 (throw (ex-info "Variation archive id was null" {:cause msg})))
-                               (log/debug :msg (format "Evidence line %s has reverse relation to %s"
-                                                       evidence-line-id variation-archive-iri))
-                               variation-archive-iri)}]}
-         }
-        ))))
+       "@reverse" {:sepio/has-evidence-line
+                   [{"@id" (let [variation-archive-iri (format (str iri/variation-archive "%s")
+                                                               (:variation_archive_id msg))]
+                             (if (empty? (:variation_archive_id msg))
+                               (throw (ex-info "Variation archive id was null" {:cause msg})))
+                             (log/debug :msg (format "Evidence line %s has reverse relation to %s"
+                                                     evidence-line-id variation-archive-iri))
+                             variation-archive-iri)}]}}))))
 
 (defmethod clinvar-model-to-jsonld :clinical_assertion [msg]
   (clinical-assertion-to-jsonld msg))

src/genegraph/transform/clinvar/util.clj

@@ -48,7 +48,7 @@
   [a b]
   (conditional-join a b "/"))
 
-(defn in? [coll e]
+(defn in? [e coll]
   (some #(= % e) coll))
 
 (defn simplify-dollar-map [m]

src/genegraph/transform/clinvar/variation.clj

@@ -442,7 +442,7 @@ query($variation_iri:String) {
     type: __typename
     ... on CategoricalVariationDescriptor {
       label
-      object {
+      value {
         id: iri
         type: __typename
         ... on CanonicalVariation {
@@ -494,6 +494,7 @@ fragment alleleFields on Allele {
 }
 ")
 
+
 (defmethod common/clinvar-add-event-graphql :variation [event]
   (let [iri (:genegraph.annotate/iri event)]
     (assoc event :graphql-params {:query variation-descriptor-query

test/genegraph/transform/clinvar/variation_transformer_test.clj

@@ -1,8 +1,10 @@
 (ns genegraph.transform.clinvar.variation-transformer-test
   (:require [cheshire.core :as json]
+            [clojure.pprint :refer [pprint]]
             [clojure.string :as s]
             [clojure.test :refer [use-fixtures]]
-            [clojure.pprint :refer [pprint]]
+            [genegraph.annotate.serialization :as ser]
+            [genegraph.sink.event :as event]
             [genegraph.transform.clinvar.core]
             [genegraph.transform.clinvar.variation :as variation]
             [genegraph.transform.types :as xform-types]
@@ -30,7 +32,8 @@
   (map xform-types/add-model-jsonld events))
 
 (defn get-variant-messages []
-  (-> "clinvar-raw-testdata_20210412.txt"
+  (-> "clinvar-raw-filtered.txt"
+      #_"clinvar-raw-testdata_20210412.txt"
       slurp
       (clojure.string/split #"\n")
       (->> (map #(json/parse-string % true))
@@ -83,3 +86,19 @@
                                 (log/info :realized realized)
                                 realized))))))))
       (#(do (def with-combined-triples %) %))))
+
+
+(defn test-vrs-normalization-acmg73 []
+  (let [events (->> (get-variant-messages)
+                    #_(filter #(= "12610" (get-in % [:content :id])))
+                    (take 1)
+                    (map eventify)
+                    (map genegraph.transform.clinvar.core/add-parsed-value)
+                    (map #(xform-types/add-model %))
+                    (map #(event/add-to-db! %))
+                    (map #((:enter ser/add-graphql-params-interceptor) %))
+                    (map #((:enter ser/add-graphql-serialization-interceptor) %))
+                    ;; (map #(xform-types/add-event-graphql %))
+                    ;; (map #(ser/add-graphql-serialization-interceptor %))
+                    #_(map #(xform-types/add-model-jsonld %)))]
+    events))