Fix conversion of AnnotationFeature to GFF3Feature (for GFF3 export) (#…

…487) * Start adding test and possibly some fixes for #481 Conversion to GFF3Feature includes source and score and sets ID and Parent attributes. However, export to GFF from UI fails. * Fix imports * Test expanded * Process CDSs, tests extended * Temporarily skip test * Skip a few more tests temporarily * Fix checks on feature import --------- Co-authored-by: Garrett Stevens <[email protected]>
GMOD · Dec 16, 2024 · 1f5ddb4 · 1f5ddb4
1 parent 0936c87
commit 1f5ddb4
Show file tree

Hide file tree

Showing 12 changed files with 560 additions and 115 deletions.
diff --git a/packages/apollo-collaboration-server/src/export/export.service.ts b/packages/apollo-collaboration-server/src/export/export.service.ts
@@ -25,7 +25,7 @@ import {
   RefSeqDocument,
 } from '@apollo-annotation/schemas'
 import {
-  makeGFF3Feature,
+  annotationFeatureToGFF3,
   splitStringIntoChunks,
 } from '@apollo-annotation/shared'
 import gff from '@gmod/gff'
@@ -179,7 +179,7 @@ export class ExportService {
             const refSeqNames = Object.fromEntries(
               refSeqs.map((refSeq) => [refSeq._id, refSeq.name]),
             )
-            const gff3Feature = makeGFF3Feature(
+            const gff3Feature = annotationFeatureToGFF3(
               flattened as unknown as AnnotationFeatureSnapshot,
               undefined,
               refSeqNames,

diff --git a/packages/apollo-collaboration-server/src/features/features.module.ts b/packages/apollo-collaboration-server/src/features/features.module.ts
@@ -30,10 +30,10 @@ import { FeaturesService } from './features.service'
         useFactory: (connection: Connection, checksService: ChecksService) => {
           FeatureSchema.plugin(idValidator, { connection })
           const runChecksOnDocument = async (doc: FeatureDocument) => {
-            await checksService.checkFeatures([doc])
+            await checksService.checkFeatures([doc], false)
           }
           const runChecksOnDocuments = async (docs: FeatureDocument[]) => {
-            await checksService.checkFeatures(docs)
+            await checksService.checkFeatures(docs, false)
           }
           FeatureSchema.post('save', runChecksOnDocument)
           FeatureSchema.post('updateOne', runChecksOnDocument)

diff --git a/packages/apollo-shared/src/GFF3/annotationFeatureToGFF3.test.ts b/packages/apollo-shared/src/GFF3/annotationFeatureToGFF3.test.ts
@@ -0,0 +1,170 @@
+/* eslint-disable prefer-destructuring */
+/* eslint-disable @typescript-eslint/no-floating-promises */
+import { describe, it } from 'node:test'
+import { assert } from 'chai'
+import { readAnnotationFeatureSnapshot } from './gff3ToAnnotationFeature.test'
+import { annotationFeatureToGFF3 } from './annotationFeatureToGFF3'
+import { AnnotationFeatureSnapshot } from '@apollo-annotation/mst'
+
+describe('annotationFeatureToGFF3', () => {
+  it('Test mandatory columns', () => {
+    const annotationFeature = readAnnotationFeatureSnapshot(
+      'test_data/gene.json',
+    )
+    const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
+
+    assert.deepEqual(gff3Feature.seq_id, 'chr1')
+    assert.deepEqual(gff3Feature.type, 'gene')
+    assert.deepEqual(gff3Feature.start, 1000)
+    assert.deepEqual(gff3Feature.end, 9000)
+    assert.deepEqual(gff3Feature.strand, '+')
+    assert.deepEqual(gff3Feature.score, 123)
+    assert.deepEqual(gff3Feature.source, 'test_data')
+  })
+  it('Feature with no children and no gff_id has no ID attribute', () => {
+    const annotationFeature = JSON.parse(`{
+      "_id": "66d70e4ccc30b55b65e5f619",
+      "refSeq": "chr1",
+      "type": "gene",
+      "min": 999,
+      "max": 9000,
+      "strand": 1,
+      "attributes": {}
+    }`) as AnnotationFeatureSnapshot
+    const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
+    assert.isUndefined(gff3Feature.attributes?.ID)
+  })
+  it('Feature with children and no gff_id has internal _id as ID', () => {
+    const annotationFeature = JSON.parse(`{
+      "_id": "66d70e4ccc30b55b65e5f619",
+      "refSeq": "chr1",
+      "type": "gene",
+      "min": 999,
+      "max": 9000,
+      "strand": 1,
+      "attributes": {},
+      "children": {
+        "66d70e4ccc30b55b65e5f618": {
+          "_id": "66d70e4ccc30b55b65e5f618",
+          "refSeq": "chr1",
+          "type": "gene_segment",
+          "min": 1049,
+          "max": 9000,
+          "strand": 1,
+          "attributes": {}
+        }
+      }
+    }`) as AnnotationFeatureSnapshot
+    const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
+    assert.deepEqual(gff3Feature.attributes?.ID, ['66d70e4ccc30b55b65e5f619'])
+  })
+  it('Convert multiple scores', () => {
+    const annotationFeature = JSON.parse(`{
+      "_id": "66d70e4ccc30b55b65e5f619",
+      "refSeq": "chr1",
+      "type": "gene",
+      "min": 999,
+      "max": 9000,
+      "strand": 1,
+      "attributes": {
+        "gff_id": ["gene10001"],
+        "gff_score": ["123", "345"]
+      }
+    }`) as AnnotationFeatureSnapshot
+    const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
+    assert.deepEqual(gff3Feature.score, 123)
+  })
+  it('Convert invalid score', () => {
+    const annotationFeature = JSON.parse(`{
+      "_id": "66d70e4ccc30b55b65e5f619",
+      "refSeq": "chr1",
+      "type": "gene",
+      "min": 999,
+      "max": 9000,
+      "strand": 1,
+      "attributes": {
+        "gff_id": ["gene10001"],
+        "gff_score": ["xyz"]
+      }
+    }`) as AnnotationFeatureSnapshot
+    const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
+    assert.deepEqual(gff3Feature.score, null)
+  })
+  it('Convert one gene test attributes', () => {
+    const annotationFeature = readAnnotationFeatureSnapshot(
+      'test_data/gene.json',
+    )
+    const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
+
+    assert.deepEqual(gff3Feature.attributes?.Name, ['EDEN'])
+    assert.deepEqual(gff3Feature.attributes?.testid, ['t001', 't003'])
+    assert.deepEqual(gff3Feature.attributes?.ID, ['gene10001'])
+    assert.deepEqual(gff3Feature.attributes?.Ontology_term, [
+      'GO1234',
+      'GO4567',
+      'SO1234',
+    ])
+    assert.deepEqual(gff3Feature.attributes?.Alias, ['myalias'])
+    assert.deepEqual(gff3Feature.attributes?.Target, ['mytarget'])
+    assert.deepEqual(gff3Feature.attributes?.Gap, ['mygap'])
+    assert.deepEqual(gff3Feature.attributes?.Derives_from, ['myderives'])
+    assert.deepEqual(gff3Feature.attributes?.Note, ['mynote'])
+    assert.deepEqual(gff3Feature.attributes?.Dbxref, ['mydbxref'])
+    assert.deepEqual(gff3Feature.attributes?.Is_circular, ['true'])
+  })
+  it('Convert one gene test children', () => {
+    const annotationFeature = readAnnotationFeatureSnapshot(
+      'test_data/gene.json',
+    )
+    const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
+    const [children] = gff3Feature.child_features
+    const [mrna] = children
+    assert.deepEqual(mrna.type, 'mRNA')
+    assert.deepEqual(mrna.attributes?.Parent, ['gene10001'])
+
+    const [cds] = mrna.child_features[2]
+    assert.deepEqual(cds.type, 'CDS')
+    assert.deepEqual(cds.attributes?.ID, ['cds10001'])
+    assert.deepEqual(cds.attributes?.Parent, ['mRNA10001'])
+  })
+  it('Convert CDSs', () => {
+    const annotationFeature = readAnnotationFeatureSnapshot(
+      'test_data/gene.json',
+    )
+    const [gff3Feature] = annotationFeatureToGFF3(annotationFeature)
+    const [children] = gff3Feature.child_features
+    const [mrna] = children
+    const cds10001 = mrna.child_features.filter((child) => {
+      const id = child[0].attributes?.ID
+      return id !== undefined && id[0] === 'cds10001'
+    })
+    assert.deepEqual(cds10001.length, 2)
+
+    const cds1_1 = cds10001[0][0]
+    assert.deepEqual(cds1_1.attributes?.ID, ['cds10001'])
+    assert.deepEqual(cds1_1.start, 1201)
+    assert.deepEqual(cds1_1.end, 1500)
+    assert.deepEqual(cds1_1.phase, '0')
+
+    const cds1_2 = cds10001[1][0]
+    assert.deepEqual(cds1_2.attributes?.ID, ['cds10001'])
+    assert.deepEqual(cds1_2.start, 5000)
+    assert.deepEqual(cds1_2.end, 5100)
+    assert.deepEqual(cds1_2.phase, '0')
+
+    assert.deepEqual(cds1_1.child_features[0][0].attributes?.ID, [
+      'cds_region10001',
+    ])
+    assert.deepEqual(cds1_1.child_features[0][0].start, 1351)
+    assert.deepEqual(cds1_1.child_features[0][0].end, 1400)
+    assert.deepEqual(cds1_1.child_features[0][0].phase, null)
+
+    const cds10004 = mrna.child_features.filter((child) => {
+      const id = child[0].attributes?.ID
+      return id !== undefined && id[0] === 'cds10004'
+    })
+    assert.deepEqual(cds10004.length, 2)
+    const cds4_1 = cds10004[0][0]
+    assert.deepEqual(cds4_1.attributes?.ID, ['cds10004'])
+  })
+})