#29 add orcid terms, add a list of obsolete terms as convenience func…

…tions, add and adapt tests
nfdi4plants · Aug 17, 2023 · 8ecd66b · 8ecd66b
1 parent 04f5158
commit 8ecd66b
Show file tree

Hide file tree

Showing 11 changed files with 366 additions and 339 deletions.
diff --git a/src/ARCTokenization/Terms.fs b/src/ARCTokenization/Terms.fs
@@ -22,6 +22,11 @@ module InvestigationMetadata =
 
     let ontology = OboOntology.fromLines true obo
 
+    let obsoleteCvTerms = 
+        ontology.Terms
+        |> List.filter (fun t -> t.IsObsolete)
+        |> List.map (fun t -> CvTerm.create(accession = t.Id, name = t.Name, ref = "INVMSO"))
+
     let cvTerms = 
         ontology.Terms
         |> List.map (fun t -> CvTerm.create(accession = t.Id, name = t.Name, ref = "INVMSO"))
@@ -32,9 +37,14 @@ module StudyMetadata =
 
     let ontology = OboOntology.fromLines true obo
 
+    let obsoleteCvTerms = 
+        ontology.Terms
+        |> List.filter (fun t -> t.IsObsolete)
+        |> List.map (fun t -> CvTerm.create(accession = t.Id, name = t.Name, ref = "STDMSO"))
+
     let cvTerms = 
         ontology.Terms
-        |> List.map (fun t -> CvTerm.create(accession = t.Id, name = t.Name, ref ="STDMSO"))    
+        |> List.map (fun t -> CvTerm.create(accession = t.Id, name = t.Name, ref = "STDMSO"))    
 
 module AssayMetadata = 
 
@@ -44,7 +54,13 @@ module AssayMetadata =
 
     let cvTerms = 
         ontology.Terms
-        |> List.map (fun t -> CvTerm.create(accession = t.Id, name = t.Name, ref ="ASSMSO"))
+        |> List.map (fun t -> CvTerm.create(accession = t.Id, name = t.Name, ref = "ASSMSO"))
+
+    let obsoleteCvTerms = 
+        ontology.Terms
+        |> List.filter (fun t -> t.IsObsolete)
+        |> List.map (fun t -> CvTerm.create(accession = t.Id, name = t.Name, ref = "ASSMSO"))
+
 
 module StructuralTerms =
 

diff --git a/src/ARCTokenization/structural_ontologies/investigation_metadata_structural_ontology.obo b/src/ARCTokenization/structural_ontologies/investigation_metadata_structural_ontology.obo
@@ -225,6 +225,21 @@ def: ""
 relationship: part_of INVMSO:00000021 ! INVESTIGATION CONTACTS
 relationship: follows INVMSO:00000031 ! Investigation Person Roles Term Accession Number
 
+[Term]
+id: INVMSO:00000093
+name: Comment[<Investigation Person ORCID>]
+def: ""
+synonym: "Comment[Investigation Person ORCID]" EXACT []
+is_obsolete: true
+relationship: part_of INVMSO:00000021 ! INVESTIGATION CONTACTS
+
+[Term]
+id: INVMSO:00000094
+name: Comment[Investigation Person ORCID]
+def: ""
+synonym: "Comment[<Investigation Person ORCID>]" EXACT []
+relationship: part_of INVMSO:00000021 ! INVESTIGATION CONTACTS
+
 [Term]
 name: STUDY
 id: INVMSO:00000033

diff --git a/src/ARCTokenization/structural_ontologies/investigation_metadata_structural_ontology.yml b/src/ARCTokenization/structural_ontologies/investigation_metadata_structural_ontology.yml
@@ -225,6 +225,21 @@ def: ""
     relationship: part_of INVMSO:00000021 ! INVESTIGATION CONTACTS
     relationship: follows INVMSO:00000031 ! Investigation Person Roles Term Accession Number
 
+    [Term]
+    id: INVMSO:00000093
+    name: Comment[<Investigation Person ORCID>]
+    def: ""
+    synonym: "Comment[Investigation Person ORCID]" EXACT []
+    is_obsolete: true
+    relationship: part_of INVMSO:00000021 ! INVESTIGATION CONTACTS
+
+    [Term]
+    id: INVMSO:00000094
+    name: Comment[Investigation Person ORCID]
+    def: ""
+    synonym: "Comment[<Investigation Person ORCID>]" EXACT []
+    relationship: part_of INVMSO:00000021 ! INVESTIGATION CONTACTS
+
   [Term]
   name: STUDY
   id: INVMSO:00000033

diff --git a/src/ARCTokenization/structural_ontologies/study_metadata_structural_ontology.obo b/src/ARCTokenization/structural_ontologies/study_metadata_structural_ontology.obo
@@ -18,7 +18,7 @@ synonym: "STUDY METADATA" EXACT []
 relationship: part_of STDMSO:00000001 ! Study Metadata
 
 [Term]
-id: STDMSO:00000051
+id: STDMSO:00000062
 name: STUDY METADATA
 def: ""
 synonym: "STUDY" EXACT []

diff --git a/src/ARCTokenization/structural_ontologies/study_metadata_structural_ontology.yml b/src/ARCTokenization/structural_ontologies/study_metadata_structural_ontology.yml
@@ -18,7 +18,7 @@ def: ""
   relationship: part_of STDMSO:00000001 ! Study Metadata
 
   [Term]
-  id: STDMSO:00000051
+  id: STDMSO:00000062
   name: STUDY METADATA
   def: ""
   synonym: "STUDY" EXACT []

diff --git a/tests/ARCTokenization.Tests/ARCTokenization.Tests.fsproj b/tests/ARCTokenization.Tests/ARCTokenization.Tests.fsproj
@@ -10,11 +10,12 @@
     <None Include="Fixtures\**" CopyToOutputDirectory="Always" />
     <Compile Include="TestUtils.fs" />
     <Compile Include="TestObjects.fs" />
+    <Compile Include="StructuralOntologytests.fs" />
+    <Compile Include="IntegrationTests\InvestigationMetadata.fs" />
     <Compile Include="IntegrationTests\StudyMetadata.fs" />
     <Compile Include="IntegrationTests\StudyAnnotationTable.fs" />
     <Compile Include="IntegrationTests\AssayMetadata.fs" />
     <Compile Include="IntegrationTests\AssayAnnotationTable.fs" />
-    <Compile Include="IntegrationTests\InvestigationMetadata.fs" />
     <Compile Include="Program.fs" />
   </ItemGroup>
 

diff --git a/tests/ARCTokenization.Tests/IntegrationTests/AssayAnnotationTable.fs b/tests/ARCTokenization.Tests/IntegrationTests/AssayAnnotationTable.fs
@@ -1,4 +1,4 @@
-module IntegrationTests.Assay
+namespace IntegrationTests
 
 open ControlledVocabulary
 open FsSpreadsheet
@@ -9,121 +9,122 @@ open Xunit
 open TestUtils
 open TestObjects
 
-module Correct  =
+module Assay = 
+    module Correct  =
 
-    module ``Assay with only source and sample column`` = 
+        module ``Assay with only source and sample column`` = 
 
-        let assay = Assays.Correct.``assay with only source and sample column``
+            let assay = Assays.Correct.``assay with only source and sample column``
 
-        [<Fact>]
-        let ``AnnotationTable count`` () =
-            Assert.Equal(assay.Length, 1) 
+            [<Fact>]
+            let ``AnnotationTable count`` () =
+                Assert.Equal(assay.Length, 1) 
 
-        let table = assay.[0] |> snd
+            let table = assay.[0] |> snd
 
-        [<Fact>]
-        let ``IOColumns count`` () =
-            TokenizedAnnotationTable.hasIOColumnAmount 2 table
+            [<Fact>]
+            let ``IOColumns count`` () =
+                TokenizedAnnotationTable.hasIOColumnAmount 2 table
 
-        [<Fact>]
-        let ``TermRelatedBuildingBlocks count`` () =
-            TokenizedAnnotationTable.hasTermRelatedBuildingBlockAmount 0 table
+            [<Fact>]
+            let ``TermRelatedBuildingBlocks count`` () =
+                TokenizedAnnotationTable.hasTermRelatedBuildingBlockAmount 0 table
 
-        let expectedIOColumns = 
-            [
+            let expectedIOColumns = 
                 [
-                    CvParam(
-                        id = "(n/a)",
-                        name = "Source Name",
-                        ref = "(n/a)",
-                        pv = (ParamValue.Value "Source A"),
-                        attributes = []
-                    )
+                    [
+                        CvParam(
+                            id = "(n/a)",
+                            name = "Source Name",
+                            ref = "(n/a)",
+                            pv = (ParamValue.Value "Source A"),
+                            attributes = []
+                        )
+                    ]
+                    [
+                        CvParam(
+                            id = "(n/a)",
+                            name = "Sample Name",
+                            ref = "(n/a)",
+                            pv = (ParamValue.Value "Sample A"),
+                            attributes = []
+                        )
+                    ]
                 ]
-                [
-                    CvParam(
-                        id = "(n/a)",
-                        name = "Sample Name",
-                        ref = "(n/a)",
-                        pv = (ParamValue.Value "Sample A"),
-                        attributes = []
-                    )
-                ]
-            ]
 
-        let expectedTermRelatedBuildingBlocks: CvParam list list = []
+            let expectedTermRelatedBuildingBlocks: CvParam list list = []
 
-        [<Fact>]
-        let ``IOColumns CvParams`` () = 
-            table 
-            |> TokenizedAnnotationTable.IOColumnsEqual expectedIOColumns
+            [<Fact>]
+            let ``IOColumns CvParams`` () = 
+                table 
+                |> TokenizedAnnotationTable.IOColumnsEqual expectedIOColumns
 
 
-        [<Fact>]
-        let ``TermRelatedBuildingBlocks CvParams`` () =
-            table 
-            |> TokenizedAnnotationTable.termRelatedBuildingBlocksEqual expectedTermRelatedBuildingBlocks
+            [<Fact>]
+            let ``TermRelatedBuildingBlocks CvParams`` () =
+                table 
+                |> TokenizedAnnotationTable.termRelatedBuildingBlocksEqual expectedTermRelatedBuildingBlocks
 
 
-    module ``Assay with single characteristics`` =
+        module ``Assay with single characteristics`` =
 
-        let assay = Assays.Correct.``assay with single characteristics``
+            let assay = Assays.Correct.``assay with single characteristics``
 
-        [<Fact>]
-        let ``AnnotationTable count`` () =
-            Assert.Equal(assay.Length, 1)
+            [<Fact>]
+            let ``AnnotationTable count`` () =
+                Assert.Equal(assay.Length, 1)
 
-        let table = assay.[0] |> snd
+            let table = assay.[0] |> snd
 
-        [<Fact>]
-        let ``IOColumns count`` () =
-            TokenizedAnnotationTable.hasIOColumnAmount 2 table
+            [<Fact>]
+            let ``IOColumns count`` () =
+                TokenizedAnnotationTable.hasIOColumnAmount 2 table
 
-        [<Fact>]
-        let ``TermRelatedBuildingBlocks count`` () =
-            TokenizedAnnotationTable.hasTermRelatedBuildingBlockAmount 1 table
+            [<Fact>]
+            let ``TermRelatedBuildingBlocks count`` () =
+                TokenizedAnnotationTable.hasTermRelatedBuildingBlockAmount 1 table
 
-        let expectedIOColumns = 
-            [
-                [
-                    CvParam(
-                        id = "(n/a)",
-                        name = "Source Name",
-                        ref = "(n/a)",
-                        pv = (ParamValue.Value "Source A"),
-                        attributes = []
-                    )
-                ]
+            let expectedIOColumns = 
                 [
-                    CvParam(
-                        id = "(n/a)",
-                        name = "Sample Name",
-                        ref = "(n/a)",
-                        pv = (ParamValue.Value "Sample A"),
-                        attributes = []
-                    )
+                    [
+                        CvParam(
+                            id = "(n/a)",
+                            name = "Source Name",
+                            ref = "(n/a)",
+                            pv = (ParamValue.Value "Source A"),
+                            attributes = []
+                        )
+                    ]
+                    [
+                        CvParam(
+                            id = "(n/a)",
+                            name = "Sample Name",
+                            ref = "(n/a)",
+                            pv = (ParamValue.Value "Sample A"),
+                            attributes = []
+                        )
+                    ]
                 ]
-            ]
 
-        let expectedTermRelatedBuildingBlocks = 
-            [
+            let expectedTermRelatedBuildingBlocks = 
                 [
-                    CvParam(
-                        id = "Term Accession Number (OBI:0100026)",
-                        name = "Characteristic [organism]",
-                        ref = "Term Source REF (OBI:0100026)",
-                        pv = ParamValue.CvValue (CvTerm.create(accession = "http://purl.obolibrary.org/obo/NCBITaxon_3702", name = "Arabidopsis thaliana", ref = "NCBITaxon")),
-                        attributes = []
-                    )
+                    [
+                        CvParam(
+                            id = "Term Accession Number (OBI:0100026)",
+                            name = "Characteristic [organism]",
+                            ref = "Term Source REF (OBI:0100026)",
+                            pv = (ParamValue.CvValue (CvTerm.create(accession = "http://purl.obolibrary.org/obo/NCBITaxon_3702", name = "Arabidopsis thaliana", ref = "NCBITaxon"))),
+                            attributes = []
+                        )
+                    ]
                 ]
-            ]
 
-        [<Fact>]
-        let ``IOColumns CvParams`` () = 
-            table 
-            |> TokenizedAnnotationTable.IOColumnsEqual expectedIOColumns
+            [<Fact>]
+            let ``IOColumns CvParams`` () = 
+                table 
+                |> TokenizedAnnotationTable.IOColumnsEqual expectedIOColumns
 
-        [<Fact>]
-        let ``TermRelatedBuildingBlocks CvParams`` () =
-            table 
-            |> TokenizedAnnotationTable.termRelatedBuildingBlocksEqual expectedTermRelatedBuildingBlocks
+            [<Fact>]
+            let ``TermRelatedBuildingBlocks CvParams`` () =
+                table 
+                |> TokenizedAnnotationTable.termRelatedBuildingBlocksEqual expectedTermRelatedBuildingBlocks
diff --git a/tests/ARCTokenization.Tests/IntegrationTests/AssayMetadata.fs b/tests/ARCTokenization.Tests/IntegrationTests/AssayMetadata.fs
@@ -1,17 +1,16 @@
-module IntegrationTests.AssayMetadata
+namespace IntegrationTests
 
-open ControlledVocabulary
-open FsSpreadsheet
-open FsSpreadsheet.ExcelIO
-open ARCTokenization
-open Xunit
+module AssayMetadata =
 
-open TestUtils
+    open ControlledVocabulary
+    open FsSpreadsheet
+    open FsSpreadsheet.ExcelIO
+    open ARCTokenization
+    open Xunit
 
+    open TestUtils
 
-
-let allExpectedMetadataTerms = 
-
-    Terms.AssayMetadata.cvTerms
-    |> List.skip 1 //(ignore root term)
-    |> List.map (fun p -> CvParam(p, ParamValue.Value "", []))
+    let allExpectedMetadataTermsEmpty = 
+        Terms.AssayMetadata.cvTerms
+        |> List.skip 1 //(ignore root term)
+        |> List.map (fun p -> CvParam(p, ParamValue.CvValue (CvTerm.create(accession = "AGMO:00000001", name = "Metadata Section Key", ref = "AGMO")), []))