Add configurable ADDITIONAL_IDS

Author: garrettjstevens

packages/apollo-collaboration-server/src/app.module.ts

@@ -57,6 +57,7 @@ const validationSchema = Joi.object({
   DESCRIPTION: Joi.string(),
   FEATURE_TYPE_ONTOLOGY_LOCATION: Joi.string(),
   PLUGIN_LOCATION: Joi.string(),
+  ADDITIONAL_IDS: Joi.string(),
   ALLOW_ROOT_USER: Joi.boolean().default(false),
   ROOT_USER_PASSWORD: Joi.string(),
   ROOT_USER_PASSWORD_FILE: Joi.string(),

packages/apollo-common/src/AssemblySpecificChange.ts

@@ -1,4 +1,6 @@
 /* eslint-disable @typescript-eslint/no-unnecessary-condition */
+import { type AnnotationFeatureSnapshot } from '@apollo-annotation/mst'
+
 import {
   Change,
   type ChangeOptions,
@@ -26,4 +28,27 @@ export abstract class AssemblySpecificChange extends Change {
     super(json, options)
     this.assembly = json.assembly
   }
+
+  getIndexedIds(
+    feature: AnnotationFeatureSnapshot,
+    additionalIds: string[] | undefined,
+  ): string[] {
+    const indexedIds: string[] = []
+    if (feature.featureId) {
+      indexedIds.push(feature.featureId)
+    }
+    for (const additionalId of additionalIds ?? []) {
+      const idValue = feature.attributes?.[additionalId]
+      if (idValue) {
+        indexedIds.push(idValue[0])
+      }
+    }
+    if (feature.children) {
+      for (const child of Object.values(feature.children)) {
+        const childIndexedIds = this.getIndexedIds(child, additionalIds)
+        indexedIds.push(...childIndexedIds)
+      }
+    }
+    return indexedIds
+  }
 }

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

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

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

@@ -89,6 +89,12 @@ export class AddFeatureChange extends FeatureChange {
     let featureCnt = 0
     logger.debug?.(`changes: ${JSON.stringify(changes)}`)
 
+    const { ADDITIONAL_IDS } = process.env
+    let additionalIds: string[] | undefined
+    if (ADDITIONAL_IDS) {
+      additionalIds = ADDITIONAL_IDS.split(',')
+    }
+
     // Loop the changes
     for (const change of changes) {
       logger.debug?.(`change: ${JSON.stringify(change)}`)
@@ -106,9 +112,10 @@ export class AddFeatureChange extends FeatureChange {
 
       // CopyFeature is called from CopyFeature.tsx
       if (copyFeature) {
+        const indexedIds = this.getIndexedIds(addedFeature, additionalIds)
         // Add into Mongo
         const [newFeatureDoc] = await featureModel.create(
-          [{ ...addedFeature, allIds, status: -1, user }],
+          [{ ...addedFeature, allIds, indexedIds, status: -1, user }],
           { session },
         )
         logger.debug?.(
@@ -143,8 +150,9 @@ export class AddFeatureChange extends FeatureChange {
         } else {
           const childIds = this.getChildFeatureIds(addedFeature)
           const allIdsV2 = [_id, ...childIds]
+          const indexedIds = this.getIndexedIds(addedFeature, additionalIds)
           const [newFeatureDoc] = await featureModel.create(
-            [{ allIds: allIdsV2, status: 0, ...addedFeature }],
+            [{ allIds: allIdsV2, indexedIds, status: 0, ...addedFeature }],
             { session },
           )
           logger.verbose?.(`Added docId "${newFeatureDoc._id}"`)

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

@@ -184,8 +184,13 @@ export abstract class FromFileBaseChange extends AssemblySpecificChange {
   }
 
   async addFeatureIntoDb(gff3Feature: GFF3Feature, backend: ServerDataStore) {
-    const { featureModel, refSeqModel, user } = backend
+    const { ADDITIONAL_IDS } = process.env
+    let additionalIds: string[] | undefined
+    if (ADDITIONAL_IDS) {
+      additionalIds = ADDITIONAL_IDS.split(',')
+    }
     const { assembly, refSeqCache } = this
+    const { featureModel, refSeqModel, user } = backend
 
     const [{ seq_id: refName }] = gff3Feature
     if (!refName) {
@@ -209,10 +214,13 @@ export abstract class FromFileBaseChange extends AssemblySpecificChange {
     }
     const newFeature = gff3ToAnnotationFeature(gff3Feature, refSeqDoc._id)
     const allIds = this.getAllIds(newFeature)
+    const indexedIds = this.getIndexedIds(newFeature, additionalIds)
 
     // Add into Mongo
     // We cannot use Mongo 'session' / transaction here because Mongo has 16 MB limit for transaction
-    await featureModel.create([{ allIds, ...newFeature, user, status: -1 }])
+    await featureModel.create([
+      { allIds, indexedIds, ...newFeature, user, status: -1 },
+    ])
   }
 
   getAllIds(feature: AnnotationFeatureSnapshot): string[] {

packages/website/docs/02-installation/04-configuration-options.md

@@ -64,6 +64,10 @@ MICROSOFT_CLIENT_SECRET=client_secret_here
 # Defaults to relative URL 'sequence_ontology.json'
 # FEATURE_TYPE_ONTOLOGY_LOCATION=sequence_ontology.json
 
+# Comma-separated list of additional attributes in features to treat as ids
+# These will be added to feature documents' "indexedIds"
+# ADDITIONAL_IDS=gene_id,transcript_id,exon_id
+
 # Application port, defaults to 3999
 # PORT=3999