Handle text values in the reference column beyond DOIs

hubmapconsortium · Aug 30, 2024 · a4dfd69 · a4dfd69
1 parent 079af45
commit a4dfd69
Show file tree

Hide file tree

Showing 3 changed files with 42 additions and 24 deletions.
diff --git a/schemas/src/digital-objects/asct-b.yaml b/schemas/src/digital-objects/asct-b.yaml
@@ -73,7 +73,7 @@ classes:
         {% for s in has_characterizing_marker_set %}
         SubClassOf(
           {% for r in s.references %}
-            Annotation(dct:reference {{r}})
+            Annotation(dct:reference "{{r}}")
           {% endfor %}
           {{id}}
           ObjectSomeValuesFrom(
@@ -105,6 +105,9 @@ classes:
     slots:
       - members
       - references
+    slot_usage:
+      references:
+        range: string
 
   AsctbRecord:
     mixins:
@@ -122,6 +125,7 @@ classes:
       - references
     slot_usage:
       references:
+        range: string
         annotations:
           owl: AnnotationAssertion, AnnotationProperty
 
@@ -168,6 +172,7 @@ classes:
       - derived_from
     slot_usage:
       references:
+        range: string
         annotations:
           owl: AnnotationAssertion, AnnotationProperty
       derived_from:
@@ -310,23 +315,23 @@ slots:
     annotations:
       owl: AnnotationAssertion, AnnotationProperty
   anatomical_structure_list:
-    required: true
+    required: false
     multivalued: true
     inlined_as_list: true
     range: AnatomicalStructureInstance
     slot_uri: ccf:anatomical_structure
     annotations:
       owl: AnnotationAssertion, AnnotationProperty
   cell_type_list:
-    required: true
+    required: false
     multivalued: true
     inlined_as_list: true
     range: CellTypeInstance
     slot_uri: ccf:cell_type
     annotations:
       owl: AnnotationAssertion, AnnotationProperty
   gene_marker_list:
-    required: true
+    required: false
     multivalued: true
     inlined_as_list: true
     range: BiomarkerInstance

diff --git a/src/normalization/normalize-asct-b.js b/src/normalization/normalize-asct-b.js
@@ -14,6 +14,7 @@ import {
   isCtIdValid,
   isIdValid,
   normalizeDoi,
+  normalizeString
 } from './patches.js';
 import { normalizeMetadata, readMetadata, writeNormalizedData, writeNormalizedMetadata } from './utils.js';
 
@@ -155,8 +156,8 @@ function normalizeCtData(context, data) {
     // Get the references
     const references = row.references
       .map((ref) => {
-        const refId = checkNotEmpty(ref.id) ? ref.id : "N/A";
-        return checkIsDoi(refId) ? normalizeDoi(refId) : refId;
+        const refString = checkNotEmpty(ref.id) ? ref.id : "N/A";
+        return checkIsDoi(refString) ? normalizeDoi(refString) : normalizeString(refString);
       });
 
     // Get the last cell type as the primary cell
@@ -283,8 +284,8 @@ function normalizeAsctbRecord(context, data) {
     // Populate all valid references
     const references = row.references
       .map((ref) => {
-        const refId = checkNotEmpty(ref.id) ? ref.id : "N/A";
-        return checkIsDoi(refId) ? normalizeDoi(refId) : refId;
+        const refString = checkNotEmpty(ref.id) ? ref.id : "N/A";
+        return checkIsDoi(refString) ? normalizeDoi(refString) : normalizeString(refString);
       });
 
     // Collect all the items
@@ -337,28 +338,31 @@ function normalizeCellMarkerDescriptor(context, data) {
     // Populate all valid references
     const references = row.references
       .map((ref) => {
-        const refId = checkNotEmpty(ref.id) ? ref.id : "N/A";
-        return checkIsDoi(refId) ? normalizeDoi(refId) : refId;
+        const refString = checkNotEmpty(ref.id) ? ref.id : "N/A";
+        return checkIsDoi(refString) ? normalizeDoi(refString) : normalizeString(refString);
       });
 
-    // Collect all the items
-    collector.push({
-      id: generateCellMarkerDescriptorId(context, recordNumber),
-      label: `Cell marker descriptor for ${primaryCt.name}`,
-      type_of: [`CellMarkerDescriptor`],
-      primary_cell_type: primaryCt.id,
-      primary_anatomical_structure: primaryAs,
-      biomarker_set: biomarkers,
-      references: references,
-      derived_from: generateAsctbRecordId(context, recordNumber)
-    });
+    // Collect all the items if the components are complete.
+    if (primaryAs && primaryCt && biomarkers) {
+      collector.push({
+        id: generateCellMarkerDescriptorId(context, recordNumber),
+        label: `Cell marker descriptor for ${primaryCt.name}`,
+        type_of: [`CellMarkerDescriptor`],
+        primary_cell_type: primaryCt.id,
+        primary_anatomical_structure: primaryAs,
+        biomarker_set: biomarkers,
+        references: references,
+        derived_from: generateAsctbRecordId(context, recordNumber)
+      });
+    }
     return collector;
   }, []);
 }
 
 function generateAsInstance(context, recordNumber, data, index) {
   const { name: doName } = context.selectedDigitalObject
-  const { id, name: asName } = data;
+  const { id, name } = data;
+  const asName = normalizeString(name);
   const orderNumber = index + 1
   return {
     id: generateAsInstanceId(context, recordNumber, orderNumber),
@@ -373,7 +377,8 @@ function generateAsInstance(context, recordNumber, data, index) {
 
 function generateCtInstance(context, recordNumber, data, index) {
   const { name: doName } = context.selectedDigitalObject
-  const { id, name: ctName } = data;
+  const { id, name } = data;
+  const ctName = normalizeString(name);
   const orderNumber = index + 1
   return {
     id: generateCtInstanceId(context, recordNumber, orderNumber),
@@ -388,7 +393,8 @@ function generateCtInstance(context, recordNumber, data, index) {
 
 function generateBmInstance(context, recordNumber, data, index) {
   const { name: doName } = context.selectedDigitalObject
-  const { id, name: bmName, b_type } = data;
+  const { id, name, b_type } = data;
+  const bmName = normalizeString(name);
   const orderNumber = index + 1
   return {
     id: generateBmInstanceId(context, recordNumber, orderNumber),

diff --git a/src/normalization/patches.js b/src/normalization/patches.js
@@ -325,6 +325,13 @@ export function normalizeDoi(doi) {
   return normDoi;
 }
 
+export function normalizeString(str) {
+  return str.replace(/"/g, "'")  // Replace all double quotes with single quotes
+    .replace(/\r/g, "")  // Remove carriage return characters
+    .replace(/\s+/g, " ")  // Replace multiple spaces (including leading/trailing spaces) with a single space
+    .trim();
+}
+
 export function isIdValid(id) {
   return /(UBERON|FMA|CL|PCL|LMHA|HGNC):\d+|https\:\/\/purl.org\/ccf\/ASCTB\-TEMP\_[a-zA-Z0-9\-]+/.test(id);
 }