Add a featureId to each annotation feature

packages/apollo-cli/src/test/test.ts

@@ -339,13 +339,10 @@ void describe('Test CLI', () => {
     assert.deepStrictEqual(p.stdout.trim(), '[]') // No failing check
 
     // Get the ID of the CDS. We need need it to modify the CDS coordinates
-    const gene = out.filter(
-      (x: any) =>
-        JSON.stringify(x.attributes.gff_id) === JSON.stringify(['gene01']),
-    )
+    const gene = out.filter((x: any) => x.featureId === 'gene01')
     const mrna = Object.values(gene.at(0).children).at(0) as any
     const cds = Object.values(mrna.children).find(
-      (x: any) => x.attributes.gff_id.at(0) === 'cds01',
+      (x: any) => x.featureId === 'cds01',
     ) as any
     const cds_id = cds._id
 
@@ -377,12 +374,11 @@ void describe('Test CLI', () => {
 
     // Get the ID of the CDS. We need need it to modify the CDS coordinates
     const gene = out.filter(
-      (x: any) =>
-        JSON.stringify(x.attributes.gff_id) === JSON.stringify(['gene01']),
+      (x: any) => x.featureId === JSON.stringify(['gene01']),
     )
     const mrna = Object.values(gene.at(0).children).at(0) as any
     const cds = Object.values(mrna.children).find(
-      (x: any) => x.attributes.gff_id.at(0) === 'cds01',
+      (x: any) => x.featureId === 'cds01',
     ) as any
     const cds_id = cds._id
 
@@ -1660,23 +1656,13 @@ void describe('Test CLI', () => {
         if (!child.attributes) {
           throw new Error('Error getting attributes')
         }
-        if (
-          JSON.stringify(child.attributes.gff_id) ===
-            JSON.stringify(['EDEN.1']) ||
-          JSON.stringify(child.attributes.gff_id) ===
-            JSON.stringify(['EDEN2.1'])
-        ) {
+        if (child.featureId === 'EDEN.1' || child.featureId === 'EDEN2.1') {
           okMrnaId.push(child._id)
         }
-        if (
-          JSON.stringify(child.attributes.gff_id) === JSON.stringify(['EDEN.2'])
-        ) {
+        if (child.featureId === 'EDEN.2') {
           warnMrnaIdForw = child._id
         }
-        if (
-          JSON.stringify(child.attributes.gff_id) ===
-          JSON.stringify(['EDEN2.2'])
-        ) {
+        if (child.featureId === 'EDEN2.2') {
           warnMrnaIdRev = child._id
         }
       }

packages/apollo-cli/tsconfig.json

@@ -10,6 +10,7 @@
     "skipLibCheck": true,
   },
   "include": ["./src"],
+  "references": [{ "path": "../apollo-mst" }, { "path": "../apollo-shared" }],
   "ts-node": {
     "esm": true,
   },

packages/apollo-mst/src/AnnotationFeatureModel.ts

@@ -41,6 +41,11 @@ type TranscriptParts = TranscriptPart[]
 export const AnnotationFeatureModel = types
   .model('AnnotationFeatureModel', {
     _id: types.identifier,
+    /**
+     * User-facing identifier for this feature (where _id is the internal
+     * identifier)
+     */
+    featureId: types.maybe(types.string),
     /** Unique ID of the reference sequence on which this feature is located */
     refSeq: types.string,
     /**
@@ -79,9 +84,6 @@ export const AnnotationFeatureModel = types
     get length() {
       return self.max - self.min
     },
-    get featureId() {
-      return self.attributes.get('id')
-    },
     /**
      * Possibly different from `min` because "The GFF3 format does not enforce a
      * rule in which features must be wholly contained within the location of

packages/apollo-schemas/src/feature.schema.ts

@@ -25,6 +25,9 @@ export class Feature
   @Prop({ type: [String], required: true, index: true })
   allIds: string[]
 
+  @Prop()
+  featureId?: string
+
   @Prop({ required: true })
   type: string
 

packages/apollo-shared/src/Changes/FromFileBaseChange.ts

@@ -4,6 +4,7 @@ import {
   AssemblySpecificChange,
   type ServerDataStore,
 } from '@apollo-annotation/common'
+import { type AnnotationFeatureSnapshot } from '@apollo-annotation/mst'
 import {
   type FileDocument,
   type RefSeqDocument,
@@ -206,19 +207,22 @@ export abstract class FromFileBaseChange extends AssemblySpecificChange {
         `RefSeq was not found by assembly "${assembly}" and seq_id "${refName}" not found`,
       )
     }
-    // Let's add featureId to parent feature
-    const featureIds: string[] = []
-
-    const newFeature = gff3ToAnnotationFeature(
-      gff3Feature,
-      refSeqDoc._id,
-      featureIds,
-    )
+    const newFeature = gff3ToAnnotationFeature(gff3Feature, refSeqDoc._id)
+    const allIds = this.getAllIds(newFeature)
 
     // Add into Mongo
     // We cannot use Mongo 'session' / transaction here because Mongo has 16 MB limit for transaction
-    await featureModel.create([
-      { allIds: featureIds, ...newFeature, user, status: -1 },
-    ])
+    await featureModel.create([{ allIds, ...newFeature, user, status: -1 }])
+  }
+
+  getAllIds(feature: AnnotationFeatureSnapshot): string[] {
+    const allIds = [feature._id]
+    if (feature.children) {
+      for (const child of Object.values(feature.children)) {
+        const childIds = this.getAllIds(child)
+        allIds.push(...childIds)
+      }
+    }
+    return allIds
   }
 }

packages/apollo-shared/src/Changes/MergeExonsChange.ts

@@ -166,36 +166,4 @@ export class MergeExonsChange extends FeatureChange {
       { logger },
     )
   }
-
-  // mergeAttributes(
-  //   firstExon: AnnotationFeatureSnapshot,
-  //   secondExon: AnnotationFeatureSnapshot,
-  // ): Record<string, string[]> {
-  //   let mergedAttrs: Record<string, string[]> = {}
-  //   if (firstExon.attributes) {
-  //     // eslint-disable-next-line unicorn/prefer-structured-clone
-  //     mergedAttrs = JSON.parse(JSON.stringify(firstExon.attributes))
-  //   }
-
-  //   if (secondExon.attributes) {
-  //     // eslint-disable-next-line unicorn/prefer-structured-clone
-  //     const attrs: Record<string, string[]> = JSON.parse(
-  //       JSON.stringify(secondExon.attributes),
-  //     )
-  //     for (const key of Object.keys(attrs)) {
-  //       if (key === '_id' || key === 'gff_id') {
-  //         continue
-  //       }
-  //       if (!Object.keys(mergedAttrs).includes(key)) {
-  //         mergedAttrs[key] = []
-  //       }
-  //       attrs[key].map((x) => {
-  //         if (!mergedAttrs[key].includes(x)) {
-  //           mergedAttrs[key].push(x)
-  //         }
-  //       })
-  //     }
-  //   }
-  //   return mergedAttrs
-  // }
 }

packages/apollo-shared/src/Changes/SplitExonChange.ts

@@ -229,8 +229,6 @@ export class SplitExonChange extends FeatureChange {
     const exon = JSON.parse(JSON.stringify(exonToBeSplit))
     // eslint-disable-next-line @typescript-eslint/no-unsafe-member-access
     delete exon.attributes._id
-    // eslint-disable-next-line @typescript-eslint/no-unsafe-member-access
-    delete exon.attributes.gff_id
 
     const leftExon = structuredClone(
       exon,

packages/apollo-shared/src/Checks/util.ts

@@ -1,9 +1,9 @@
 import { type AnnotationFeatureSnapshot } from '@apollo-annotation/mst'
 
 export function getPrintableId(feature: AnnotationFeatureSnapshot): string {
-  const gff_id = feature.attributes?.gff_id?.join(', ')
-  if (gff_id) {
-    return `${gff_id} (_id: ${feature._id.toString()})`
+  const { featureId } = feature
+  if (featureId) {
+    return `${featureId} (_id: ${feature._id.toString()})`
   }
   const gff_name = feature.attributes?.gff_name?.join(', ')
   if (gff_name) {

packages/apollo-shared/src/GFF3/annotationFeatureToGFF3.test.ts

@@ -1,12 +1,34 @@
 /* eslint-disable prefer-destructuring */
 /* eslint-disable @typescript-eslint/no-floating-promises */
+import { readFileSync } from 'node:fs'
 import { describe, it } from 'node:test'
 
 import { type AnnotationFeatureSnapshot } from '@apollo-annotation/mst'
-import { assert } from 'chai'
+import { formatSync } from '@gmod/gff'
+import { assert, expect } from 'chai'
 
 import { annotationFeatureToGFF3 } from './annotationFeatureToGFF3'
-import { readAnnotationFeatureSnapshot } from './gff3ToAnnotationFeature.test'
+import { readAnnotationFeatureSnapshot, testCases } from './testUtil'
+
+describe('Converts AnnotationFeatureSnapshot JSON to GFF3 when', () => {
+  for (const testCase of testCases) {
+    const { filenameStem, description } = testCase
+    it(description, () => {
+      const annotationFeatures = JSON.parse(
+        readFileSync(`test_data/${filenameStem}.json`, 'utf8'),
+      ) as AnnotationFeatureSnapshot[]
+      const expectedGFF3 = readFileSync(
+        `test_data/${filenameStem}.gff3`,
+        'utf8',
+      )
+      const gffFeatures = annotationFeatures.map((annotationFeature) =>
+        annotationFeatureToGFF3(annotationFeature),
+      )
+      const gff3 = formatSync(gffFeatures)
+      expect(gff3).to.equal(expectedGFF3)
+    })
+  }
+})
 
 describe('annotationFeatureToGFF3', () => {
   it('Test mandatory columns', () => {
@@ -69,9 +91,9 @@ describe('annotationFeatureToGFF3', () => {
       "max": 9000,
       "strand": 1,
       "attributes": {
-        "gff_id": ["gene10001"],
         "gff_score": ["123", "345"]
-      }
+      },
+      "featureId": "gene10001"
     }`) as AnnotationFeatureSnapshot
     const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
     assert.deepEqual(gff3Feature.score, 123)
@@ -85,9 +107,9 @@ describe('annotationFeatureToGFF3', () => {
       "max": 9000,
       "strand": 1,
       "attributes": {
-        "gff_id": ["gene10001"],
         "gff_score": ["xyz"]
-      }
+      },
+      "featureId": "gene10001"
     }`) as AnnotationFeatureSnapshot
     const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
     assert.deepEqual(gff3Feature.score, null)

packages/apollo-shared/src/GFF3/annotationFeatureToGFF3.ts

@@ -24,10 +24,13 @@ export function annotationFeatureToGFF3(
   if (parentId) {
     attributes.Parent = [parentId]
   }
+  if (feature.featureId) {
+    attributes.ID = [feature.featureId]
+  }
   if (attributes.gff_id) {
-    attributes.ID = attributes.gff_id
     delete attributes.gff_id
-  } else if (feature.children) {
+  }
+  if (feature.children && !attributes.ID) {
     attributes.ID = [feature._id]
   }
   if (attributes.gff_name) {