@apollo-annotation/shared 0.2.0 → 0.2.2

package/dist/util.d.ts

@@ -1,4 +1 @@
-import { AnnotationFeatureSnapshot } from '@apollo-annotation/mst';
-import { GFF3Feature } from '@gmod/gff';
-export declare function makeGFF3Feature(feature: AnnotationFeatureSnapshot, parentId?: string, refSeqNames?: Record<string, string | undefined>): GFF3Feature;
 export declare function splitStringIntoChunks(input: string, chunkSize: number): string[];

package/dist/util.js

@@ -1,96 +1,6 @@
 "use strict";
-/* eslint-disable @typescript-eslint/no-unsafe-assignment */
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.makeGFF3Feature = makeGFF3Feature;
 exports.splitStringIntoChunks = splitStringIntoChunks;
-function makeGFF3Feature(feature, parentId, refSeqNames) {
-    const locations = [{ start: feature.min, end: feature.max }];
-    // const locations = feature.discontinuousLocations?.length
-    //   ? feature.discontinuousLocations
-    //   : [{ start: feature.start, end: feature.end, phase: feature.phase }]
-    const attributes = JSON.parse(JSON.stringify(feature.attributes));
-    const ontologyTerms = [];
-    const source = feature.attributes?.source?.[0] ?? null;
-    delete attributes.source;
-    if (parentId) {
-        attributes.Parent = [parentId];
-    }
-    if (attributes._id) {
-        attributes.ID = attributes._id;
-        delete attributes._id;
-    }
-    if (attributes.gff_name) {
-        attributes.Name = attributes.gff_name;
-        delete attributes.gff_name;
-    }
-    if (attributes.gff_alias) {
-        attributes.Alias = attributes.gff_alias;
-        delete attributes.gff_alias;
-    }
-    if (attributes.gff_target) {
-        attributes.Target = attributes.gff_target;
-        delete attributes.gff_target;
-    }
-    if (attributes.gff_gap) {
-        attributes.Gap = attributes.gff_gap;
-        delete attributes.gff_gap;
-    }
-    if (attributes.gff_derives_from) {
-        attributes.Derives_from = attributes.gff_derives_from;
-        delete attributes.gff_derives_from;
-    }
-    if (attributes.gff_note) {
-        attributes.Note = attributes.gff_note;
-        delete attributes.gff_note;
-    }
-    if (attributes.gff_dbxref) {
-        attributes.Dbxref = attributes.gff_dbxref;
-        delete attributes.gff_dbxref;
-    }
-    if (attributes.gff_is_circular) {
-        attributes.Is_circular = attributes.gff_is_circular;
-        delete attributes.gff_is_circular;
-    }
-    if (attributes.gff_ontology_term) {
-        ontologyTerms.push(...attributes.gff_ontology_term);
-        delete attributes.gff_ontology_term;
-    }
-    if (attributes['Gene Ontology']) {
-        ontologyTerms.push(...attributes['Gene Ontology']);
-        delete attributes['Gene Ontology'];
-    }
-    if (attributes['Sequence Ontology']) {
-        ontologyTerms.push(...attributes['Sequence Ontology']);
-        delete attributes['Sequence Ontology'];
-    }
-    if (ontologyTerms.length > 0) {
-        attributes.Ontology_term = ontologyTerms;
-    }
-    return locations.map((location) => ({
-        start: location.start + 1,
-        end: location.end,
-        seq_id: refSeqNames ? refSeqNames[feature.refSeq] ?? null : feature.refSeq,
-        source,
-        type: feature.type,
-        score: null,
-        // score: feature.score ?? null,
-        strand: feature.strand ? (feature.strand === 1 ? '+' : '-') : null,
-        phase: null,
-        // phase:
-        //   location.phase === 0
-        //     ? '0'
-        //     : location.phase === 1
-        //       ? '1'
-        //       : location.phase === 2
-        //         ? '2'
-        //         : null,
-        attributes: Object.keys(attributes).length > 0 ? attributes : null,
-        derived_features: [],
-        child_features: feature.children
-            ? Object.values(feature.children).map((child) => makeGFF3Feature(child, attributes.ID?.[0], refSeqNames))
-            : [],
-    }));
-}
 function splitStringIntoChunks(input, chunkSize) {
     const chunks = [];
     for (let i = 0; i < input.length; i += chunkSize) {

package/dist/util.js.map

@@ -1 +1 @@
-{"version":3,"file":"util.js","sourceRoot":"","sources":["../src/util.ts"],"names":[],"mappings":";AAAA,4DAA4D;;AAK5D,0CA+FC;AAED,sDAUC;AA3GD,SAAgB,eAAe,CAC7B,OAAkC,EAClC,QAAiB,EACjB,WAAgD;IAEhD,MAAM,SAAS,GAAG,CAAC,EAAE,KAAK,EAAE,OAAO,CAAC,GAAG,EAAE,GAAG,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC,CAAA;IAC5D,2DAA2D;IAC3D,qCAAqC;IACrC,yEAAyE;IACzE,MAAM,UAAU,GAAyC,IAAI,CAAC,KAAK,CACjE,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC,UAAU,CAAC,CACnC,CAAA;IACD,MAAM,aAAa,GAAa,EAAE,CAAA;IAClC,MAAM,MAAM,GAAG,OAAO,CAAC,UAAU,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,IAAI,IAAI,CAAA;IACtD,OAAO,UAAU,CAAC,MAAM,CAAA;IACxB,IAAI,QAAQ,EAAE,CAAC;QACb,UAAU,CAAC,MAAM,GAAG,CAAC,QAAQ,CAAC,CAAA;IAChC,CAAC;IACD,IAAI,UAAU,CAAC,GAAG,EAAE,CAAC;QACnB,UAAU,CAAC,EAAE,GAAG,UAAU,CAAC,GAAG,CAAA;QAC9B,OAAO,UAAU,CAAC,GAAG,CAAA;IACvB,CAAC;IACD,IAAI,UAAU,CAAC,QAAQ,EAAE,CAAC;QACxB,UAAU,CAAC,IAAI,GAAG,UAAU,CAAC,QAAQ,CAAA;QACrC,OAAO,UAAU,CAAC,QAAQ,CAAA;IAC5B,CAAC;IACD,IAAI,UAAU,CAAC,SAAS,EAAE,CAAC;QACzB,UAAU,CAAC,KAAK,GAAG,UAAU,CAAC,SAAS,CAAA;QACvC,OAAO,UAAU,CAAC,SAAS,CAAA;IAC7B,CAAC;IACD,IAAI,UAAU,CAAC,UAAU,EAAE,CAAC;QAC1B,UAAU,CAAC,MAAM,GAAG,UAAU,CAAC,UAAU,CAAA;QACzC,OAAO,UAAU,CAAC,UAAU,CAAA;IAC9B,CAAC;IACD,IAAI,UAAU,CAAC,OAAO,EAAE,CAAC;QACvB,UAAU,CAAC,GAAG,GAAG,UAAU,CAAC,OAAO,CAAA;QACnC,OAAO,UAAU,CAAC,OAAO,CAAA;IAC3B,CAAC;IACD,IAAI,UAAU,CAAC,gBAAgB,EAAE,CAAC;QAChC,UAAU,CAAC,YAAY,GAAG,UAAU,CAAC,gBAAgB,CAAA;QACrD,OAAO,UAAU,CAAC,gBAAgB,CAAA;IACpC,CAAC;IACD,IAAI,UAAU,CAAC,QAAQ,EAAE,CAAC;QACxB,UAAU,CAAC,IAAI,GAAG,UAAU,CAAC,QAAQ,CAAA;QACrC,OAAO,UAAU,CAAC,QAAQ,CAAA;IAC5B,CAAC;IACD,IAAI,UAAU,CAAC,UAAU,EAAE,CAAC;QAC1B,UAAU,CAAC,MAAM,GAAG,UAAU,CAAC,UAAU,CAAA;QACzC,OAAO,UAAU,CAAC,UAAU,CAAA;IAC9B,CAAC;IACD,IAAI,UAAU,CAAC,eAAe,EAAE,CAAC;QAC/B,UAAU,CAAC,WAAW,GAAG,UAAU,CAAC,eAAe,CAAA;QACnD,OAAO,UAAU,CAAC,eAAe,CAAA;IACnC,CAAC;IACD,IAAI,UAAU,CAAC,iBAAiB,EAAE,CAAC;QACjC,aAAa,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,iBAAiB,CAAC,CAAA;QACnD,OAAO,UAAU,CAAC,iBAAiB,CAAA;IACrC,CAAC;IACD,IAAI,UAAU,CAAC,eAAe,CAAC,EAAE,CAAC;QAChC,aAAa,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,eAAe,CAAC,CAAC,CAAA;QAClD,OAAO,UAAU,CAAC,eAAe,CAAC,CAAA;IACpC,CAAC;IACD,IAAI,UAAU,CAAC,mBAAmB,CAAC,EAAE,CAAC;QACpC,aAAa,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,mBAAmB,CAAC,CAAC,CAAA;QACtD,OAAO,UAAU,CAAC,mBAAmB,CAAC,CAAA;IACxC,CAAC;IACD,IAAI,aAAa,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC7B,UAAU,CAAC,aAAa,GAAG,aAAa,CAAA;IAC1C,CAAC;IACD,OAAO,SAAS,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QAClC,KAAK,EAAE,QAAQ,CAAC,KAAK,GAAG,CAAC;QACzB,GAAG,EAAE,QAAQ,CAAC,GAAG;QACjB,MAAM,EAAE,WAAW,CAAC,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM;QAC1E,MAAM;QACN,IAAI,EAAE,OAAO,CAAC,IAAI;QAClB,KAAK,EAAE,IAAI;QACX,gCAAgC;QAChC,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI;QAClE,KAAK,EAAE,IAAI;QACX,SAAS;QACT,yBAAyB;QACzB,YAAY;QACZ,6BAA6B;QAC7B,cAAc;QACd,+BAA+B;QAC/B,gBAAgB;QAChB,kBAAkB;QAClB,UAAU,EAAE,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI;QAClE,gBAAgB,EAAE,EAAE;QACpB,cAAc,EAAE,OAAO,CAAC,QAAQ;YAC9B,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAC5C,eAAe,CAAC,KAAK,EAAE,UAAU,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,WAAW,CAAC,CACxD;YACH,CAAC,CAAC,EAAE;KACP,CAAC,CAAC,CAAA;AACL,CAAC;AAED,SAAgB,qBAAqB,CACnC,KAAa,EACb,SAAiB;IAEjB,MAAM,MAAM,GAAa,EAAE,CAAA;IAC3B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,IAAI,SAAS,EAAE,CAAC;QACjD,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,CAAA;QAC3C,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;IACpB,CAAC;IACD,OAAO,MAAM,CAAA;AACf,CAAC"}
+{"version":3,"file":"util.js","sourceRoot":"","sources":["../src/util.ts"],"names":[],"mappings":";;AAAA,sDAUC;AAVD,SAAgB,qBAAqB,CACnC,KAAa,EACb,SAAiB;IAEjB,MAAM,MAAM,GAAa,EAAE,CAAA;IAC3B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,IAAI,SAAS,EAAE,CAAC;QACjD,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,CAAA;QAC3C,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;IACpB,CAAC;IACD,OAAO,MAAM,CAAA;AACf,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@apollo-annotation/shared",
-  "version": "0.2.0",
+  "version": "0.2.2",
   "main": "./dist/index.js",
   "scripts": {
     "build": "yarn clean && tsc --build",
@@ -10,9 +10,9 @@
     "test:ci": "NODE_V8_COVERAGE=./coverage glob -c \"tsx --test --test-reporter spec --experimental-test-coverage \" \"**/*.test.ts\""
   },
   "dependencies": {
-    "@apollo-annotation/common": "^0.2.0",
-    "@apollo-annotation/mst": "^0.2.0",
-    "@apollo-annotation/schemas": "^0.2.0",
+    "@apollo-annotation/common": "^0.2.2",
+    "@apollo-annotation/mst": "^0.2.2",
+    "@apollo-annotation/schemas": "^0.2.2",
     "@gmod/gff": "1.2.0",
     "@gmod/indexedfasta": "^2.0.4",
     "@jbrowse/core": "^2.13.1",
@@ -30,7 +30,7 @@
     "chai-exclude": "^3.0.0",
     "glob": "^11.0.0",
     "mobx": "^6.6.1",
-    "mobx-state-tree": "^5.1.7",
+    "mobx-state-tree": "^5.4.0",
     "mongoose": "^6.12.0",
     "react-dom": "^18.2.0",
     "rimraf": "^3.0.2",
@@ -43,7 +43,7 @@
     "@mui/x-data-grid": "^7.0.0",
     "mobx": "^6.6.1",
     "mobx-react": "^7.2.1",
-    "mobx-state-tree": "^5.1.7",
+    "mobx-state-tree": "^5.4.0",
     "prop-types": "^15.8.1",
     "react": "^18.2.0",
     "react-dom": "^18.2.0",

package/src/Checks/CDSCheck.ts

@@ -124,7 +124,7 @@ async function checkMRNA(
           refSeq: refSeq.toString(),
           start: max,
           end: max,
-          message: `Feature "${_id}" is missing a stop codon`,
+          message: `Missing stop codon`,
         })
       }
     } else {

package/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'])
+  })
+})

package/src/GFF3/annotationFeatureToGFF3.ts

@@ -0,0 +1,257 @@
+/* eslint-disable @typescript-eslint/no-unsafe-assignment */
+
+import {
+  AnnotationFeatureSnapshot,
+  TranscriptPartLocation,
+  TranscriptPartNonCoding,
+} from '@apollo-annotation/mst'
+import { GFF3Feature } from '@gmod/gff'
+import { intersection2 } from '@jbrowse/core/util'
+
+export function annotationFeatureToGFF3(
+  feature: AnnotationFeatureSnapshot,
+  parentId?: string,
+  refSeqNames?: Record<string, string | undefined>,
+): GFF3Feature {
+  const attributes: Record<string, string[] | undefined> = JSON.parse(
+    JSON.stringify(feature.attributes),
+  )
+  const ontologyTerms: string[] = []
+  const source = feature.attributes?.gff_source?.[0] ?? null
+
+  delete attributes.gff_source
+  if (parentId) {
+    attributes.Parent = [parentId]
+  }
+  if (attributes.gff_id) {
+    attributes.ID = attributes.gff_id
+    delete attributes.gff_id
+  } else if (feature.children) {
+    attributes.ID = [feature._id]
+  }
+  if (attributes.gff_name) {
+    attributes.Name = attributes.gff_name
+    delete attributes.gff_name
+  }
+  if (attributes.gff_alias) {
+    attributes.Alias = attributes.gff_alias
+    delete attributes.gff_alias
+  }
+  if (attributes.gff_target) {
+    attributes.Target = attributes.gff_target
+    delete attributes.gff_target
+  }
+  if (attributes.gff_gap) {
+    attributes.Gap = attributes.gff_gap
+    delete attributes.gff_gap
+  }
+  if (attributes.gff_derives_from) {
+    attributes.Derives_from = attributes.gff_derives_from
+    delete attributes.gff_derives_from
+  }
+  if (attributes.gff_note) {
+    attributes.Note = attributes.gff_note
+    delete attributes.gff_note
+  }
+  if (attributes.gff_dbxref) {
+    attributes.Dbxref = attributes.gff_dbxref
+    delete attributes.gff_dbxref
+  }
+  if (attributes.gff_is_circular) {
+    attributes.Is_circular = attributes.gff_is_circular
+    delete attributes.gff_is_circular
+  }
+  if (attributes.gff_ontology_term) {
+    ontologyTerms.push(...attributes.gff_ontology_term)
+    delete attributes.gff_ontology_term
+  }
+  if (attributes['Gene Ontology']) {
+    ontologyTerms.push(...attributes['Gene Ontology'])
+    delete attributes['Gene Ontology']
+  }
+  if (attributes['Sequence Ontology']) {
+    ontologyTerms.push(...attributes['Sequence Ontology'])
+    delete attributes['Sequence Ontology']
+  }
+  if (ontologyTerms.length > 0) {
+    attributes.Ontology_term = ontologyTerms
+  }
+
+  const gff_score = feature.attributes?.gff_score
+  let score: number | null = null
+  if (gff_score && gff_score.length > 0) {
+    if (gff_score[0]) {
+      score = Number(gff_score[0])
+      if (Number.isNaN(score)) {
+        score = null
+      }
+    }
+    delete attributes.gff_score
+  }
+
+  const locations = [{ start: feature.min, end: feature.max }]
+
+  return locations.map((location) => ({
+    start: Number(location.start) + 1,
+    end: Number(location.end),
+    seq_id: refSeqNames ? refSeqNames[feature.refSeq] ?? null : feature.refSeq,
+    source,
+    type: feature.type,
+    score,
+    strand: feature.strand ? (feature.strand === 1 ? '+' : '-') : null,
+    phase: null,
+    attributes: Object.keys(attributes).length > 0 ? attributes : null,
+    derived_features: [],
+    child_features: prepareChildFeatures(
+      feature,
+      attributes.ID?.[0],
+      refSeqNames,
+    ),
+  }))
+}
+
+function prepareChildFeatures(
+  feature: AnnotationFeatureSnapshot,
+  parentID?: string,
+  refSeqNames?: Record<string, string | undefined>,
+): GFF3Feature[] {
+  if (!feature.children) {
+    return []
+  }
+  if (feature.type === 'mRNA') {
+    const child_features: GFF3Feature[] = []
+    const cdsLocations = getCdsLocations(feature)
+    let cds_idx = 0
+    for (const child of Object.values(feature.children)) {
+      const gffChild = annotationFeatureToGFF3(child, parentID, refSeqNames)
+      if (child.type === 'CDS') {
+        for (const loc of cdsLocations[cds_idx]) {
+          const gffCds = JSON.parse(JSON.stringify(gffChild)) as GFF3Feature
+          if (gffCds.length != 1) {
+            // Do we need this check?
+            throw new Error(
+              `Unexpected CDS: ${JSON.stringify(gffCds, null, 2)}`,
+            )
+          }
+          gffCds[0].start = loc.min + 1
+          gffCds[0].end = loc.max
+          gffCds[0].phase = loc.phase.toString()
+          gffCds[0].type = loc.type // Do we need this?
+          child_features.push(gffCds)
+        }
+        cds_idx++
+      } else {
+        child_features.push(gffChild)
+      }
+    }
+    return child_features
+  }
+  return Object.values(feature.children).map((child) =>
+    annotationFeatureToGFF3(child, parentID, refSeqNames),
+  )
+}
+
+interface TranscriptPartCoding extends TranscriptPartLocation {
+  type: 'CDS'
+  phase: 0 | 1 | 2
+}
+type TranscriptPart = TranscriptPartCoding | TranscriptPartNonCoding
+type TranscriptParts = TranscriptPart[]
+
+function getTranscriptParts(
+  feature: AnnotationFeatureSnapshot,
+): TranscriptParts[] {
+  if (feature.type !== 'mRNA') {
+    throw new Error(
+      'Only features of type "mRNA" or equivalent can calculate CDS locations',
+    )
+  }
+  if (!feature.children) {
+    throw new Error('no CDS or exons in mRNA')
+  }
+  // In AnnotationFeatureModel we have `children.values()`
+  const children = Object.values(feature.children)
+  const cdsChildren = children.filter((child) => child.type === 'CDS')
+  if (cdsChildren.length === 0) {
+    throw new Error('no CDS in mRNA')
+  }
+  const transcriptParts: TranscriptParts[] = []
+  for (const cds of cdsChildren) {
+    const { max: cdsMax, min: cdsMin } = cds
+    const parts: TranscriptParts = []
+    let hasIntersected = false
+    const exonLocations: TranscriptPartLocation[] = []
+    for (const child of children) {
+      if (child.type === 'exon') {
+        exonLocations.push({ min: child.min, max: child.max })
+      }
+    }
+    exonLocations.sort(({ min: a }, { min: b }) => a - b)
+    for (const child of exonLocations) {
+      const lastPart = parts.at(-1)
+      if (lastPart) {
+        parts.push({ min: lastPart.max, max: child.min, type: 'intron' })
+      }
+      const [start, end] = intersection2(cdsMin, cdsMax, child.min, child.max)
+      let utrType: 'fivePrimeUTR' | 'threePrimeUTR'
+      if (hasIntersected) {
+        utrType = feature.strand === 1 ? 'threePrimeUTR' : 'fivePrimeUTR'
+      } else {
+        utrType = feature.strand === 1 ? 'fivePrimeUTR' : 'threePrimeUTR'
+      }
+      if (start !== undefined && end !== undefined) {
+        hasIntersected = true
+        if (start === child.min && end === child.max) {
+          parts.push({ min: start, max: end, phase: 0, type: 'CDS' })
+        } else if (start === child.min) {
+          parts.push(
+            { min: start, max: end, phase: 0, type: 'CDS' },
+            { min: end, max: child.max, type: utrType },
+          )
+        } else if (end === child.max) {
+          parts.push(
+            { min: child.min, max: start, type: utrType },
+            { min: start, max: end, phase: 0, type: 'CDS' },
+          )
+        } else {
+          parts.push(
+            { min: child.min, max: start, type: utrType },
+            { min: start, max: end, phase: 0, type: 'CDS' },
+            {
+              min: end,
+              max: child.max,
+              type:
+                utrType === 'fivePrimeUTR' ? 'threePrimeUTR' : 'fivePrimeUTR',
+            },
+          )
+        }
+      } else {
+        parts.push({ min: child.min, max: child.max, type: utrType })
+      }
+    }
+    parts.sort(({ min: a }, { min: b }) => a - b)
+    if (feature.strand === -1) {
+      parts.reverse()
+    }
+    let nextPhase: 0 | 1 | 2 = 0
+    const phasedParts = parts.map((loc) => {
+      if (loc.type !== 'CDS') {
+        return loc
+      }
+      const phase = nextPhase
+      nextPhase = ((3 - ((loc.max - loc.min - phase + 3) % 3)) % 3) as 0 | 1 | 2
+      return { ...loc, phase }
+    })
+    transcriptParts.push(phasedParts)
+  }
+  return transcriptParts
+}
+
+function getCdsLocations(
+  feature: AnnotationFeatureSnapshot,
+): TranscriptPartCoding[][] {
+  const transcriptParts = getTranscriptParts(feature)
+  return transcriptParts.map((transcript) =>
+    transcript.filter((transcriptPart) => transcriptPart.type === 'CDS'),
+  )
+}

package/src/GFF3/gff3ToAnnotationFeature.test.ts

@@ -120,7 +120,9 @@ function readFeatureFile(fn: string): GFF3Feature[] {
   return inGff
 }
 
-function readAnnotationFeatureSnapshot(fn: string): AnnotationFeatureSnapshot {
+export function readAnnotationFeatureSnapshot(
+  fn: string,
+): AnnotationFeatureSnapshot {
   const lines = readFileSync(fn).toString()
   return JSON.parse(lines) as AnnotationFeatureSnapshot
 }

package/src/GFF3/index.ts

@@ -1,2 +1,3 @@
+export * from './annotationFeatureToGFF3'
 export * from './gffReservedKeys'
 export * from './gff3ToAnnotationFeature'