@apollo-annotation/shared 0.3.8 → 0.3.10

package/src/GFF3/gff3ToAnnotationFeature.ts

@@ -8,7 +8,6 @@ import { gffToInternal, isGFFReservedAttribute } from './gffReservedKeys'
 export function gff3ToAnnotationFeature(
   gff3Feature: GFF3Feature,
   refSeq?: string,
-  featureIds?: string[],
 ): AnnotationFeatureSnapshot {
   const [firstFeature] = gff3Feature
   const { end, seq_id: refName, start, strand, type } = firstFeature
@@ -35,7 +34,7 @@ export function gff3ToAnnotationFeature(
 
   const [min, max] = getFeatureMinMax(gff3Feature)
 
-  const convertedChildren = convertChildren(gff3Feature, refSeq, featureIds)
+  const convertedChildren = convertChildren(gff3Feature, refSeq)
 
   const convertedAttributes = convertFeatureAttributes(gff3Feature)
 
@@ -61,9 +60,6 @@ export function gff3ToAnnotationFeature(
   if (convertedAttributes) {
     feature.attributes = convertedAttributes
   }
-  if (featureIds) {
-    featureIds.push(feature._id)
-  }
   return feature
 }
 
@@ -115,7 +111,7 @@ function convertFeatureAttributes(
   if (attributesCollections.length > 0) {
     for (const attributesCollection of attributesCollections) {
       for (const [key, val] of Object.entries(attributesCollection)) {
-        if (!val || key === 'Parent') {
+        if (key === 'Parent') {
           continue
         }
         const newKey = isGFFReservedAttribute(key) ? gffToInternal[key] : key
@@ -182,13 +178,13 @@ function convertChildren(
     if (firstChildFeatureLocation.type === 'CDS') {
       cdsFeatures.push(childFeature)
     } else {
-      const child = gff3ToAnnotationFeature(childFeature, refSeq, featureIds)
+      const child = gff3ToAnnotationFeature(childFeature, refSeq)
       convertedChildren[child._id] = child
     }
   }
 
   if (cdsFeatures.length > 0) {
-    const processedCDS = processCDS(cdsFeatures, refSeq, featureIds)
+    const processedCDS = processCDS(cdsFeatures, refSeq)
 
     for (const cds of processedCDS) {
       convertedChildren[cds._id] = cds
@@ -372,16 +368,13 @@ function mergeAnnotationFeatures(
 function processCDS(
   cdsFeatures: GFF3Feature[],
   refSeq?: string,
-  featureIds?: string[],
 ): AnnotationFeatureSnapshot[] {
   const locationCounts = cdsFeatures.map((cds) => cds.length)
   // If any CDS have multiple locations, assume it really is multiple CDS
   // (e.g. the mRNA has multiple alternative translational start sites)
   // and process normally.
   if (locationCounts.some((count) => count > 1)) {
-    return cdsFeatures.map((cds) =>
-      gff3ToAnnotationFeature(cds, refSeq, featureIds),
-    )
+    return cdsFeatures.map((cds) => gff3ToAnnotationFeature(cds, refSeq))
   }
   // If all CDS have a single location, we guess that this GFF3 represented CDS
   // as multiple features instead of a single feature with multiple locations.
@@ -402,7 +395,7 @@ function processCDS(
   })
   // If no overlaps, assume it's a single CDS feature
   if (!overlapping) {
-    return [gff3ToAnnotationFeature(sortedCDSLocations, refSeq, featureIds)]
+    return [gff3ToAnnotationFeature(sortedCDSLocations, refSeq)]
   }
   // Some CDS locations overlap, the best we can do is use the original order to
   // guess how to group the locations into features
@@ -430,7 +423,5 @@ function processCDS(
       lastGroup.push(location)
     }
   }
-  return groupedLocations.map((group) =>
-    gff3ToAnnotationFeature(group, refSeq, featureIds),
-  )
+  return groupedLocations.map((group) => gff3ToAnnotationFeature(group, refSeq))
 }

package/src/GFF3/testUtil.ts

@@ -0,0 +1,25 @@
+import { readFileSync } from 'node:fs'
+
+import { type AnnotationFeatureSnapshot } from '@apollo-annotation/mst'
+
+export function readAnnotationFeatureSnapshot(
+  fn: string,
+): AnnotationFeatureSnapshot {
+  const lines = readFileSync(fn).toString()
+  return JSON.parse(lines) as AnnotationFeatureSnapshot
+}
+
+export const testCases: { filenameStem: string; description: string }[] = [
+  {
+    filenameStem: 'single_feature_no_children',
+    description: 'there is a single feature with no children',
+  },
+  {
+    filenameStem: 'single_feature_two_children',
+    description: 'there is a single feature with two children',
+  },
+  {
+    filenameStem: 'gene_with_two_cds',
+    description: 'Gene with two CDS',
+  },
+]

package/test_data/gene_with_two_cds.gff3

@@ -0,0 +1,6 @@
+chr1	.	gene	1000	9000	.	+	.	testid=t003;ID=gene10001;Name=EDEN
+chr1	.	mRNA	1050	9000	.	+	.	testid=t004,t001,t004;Parent=gene10001;ID=mRNA10001;Name=EDEN.1
+chr1	.	exon	1050	1500	.	+	.	testid=t007;Parent=mRNA10001;ID=exon10001
+chr1	.	exon	5000	5500	.	+	.	testid=t010;Parent=mRNA10001;ID=exon10004
+chr1	.	CDS	1201	1500	.	+	0	testid=t012,t013,t014;Parent=mRNA10001;ID=cds10001;Name=edenprotein.1
+chr1	.	CDS	5000	5000	.	+	0	testid=t012,t013,t014;Parent=mRNA10001;ID=cds10001;Name=edenprotein.1

package/test_data/gene_with_two_cds.json

@@ -0,0 +1,69 @@
+[
+  {
+    "_id": "66d70f3b9c7a7460925687a3",
+    "refSeq": "chr1",
+    "type": "gene",
+    "min": 999,
+    "max": 9000,
+    "strand": 1,
+    "children": {
+      "66d70f3b9c7a7460925687a2": {
+        "_id": "66d70f3b9c7a7460925687a2",
+        "refSeq": "chr1",
+        "type": "mRNA",
+        "min": 1049,
+        "max": 9000,
+        "strand": 1,
+        "children": {
+          "66d70f3b9c7a74609256879f": {
+            "_id": "66d70f3b9c7a74609256879f",
+            "refSeq": "chr1",
+            "type": "exon",
+            "min": 1049,
+            "max": 1500,
+            "strand": 1,
+            "attributes": {
+              "testid": ["t007"],
+              "gff_id": ["exon10001"]
+            }
+          },
+          "66d70f3b9c7a7460925687a0": {
+            "_id": "66d70f3b9c7a7460925687a0",
+            "refSeq": "chr1",
+            "type": "exon",
+            "min": 4999,
+            "max": 5500,
+            "strand": 1,
+            "attributes": {
+              "testid": ["t010"],
+              "gff_id": ["exon10004"]
+            }
+          },
+          "66d70f3b9c7a7460925687a1": {
+            "_id": "66d70f3b9c7a7460925687a1",
+            "refSeq": "chr1",
+            "type": "CDS",
+            "min": 1200,
+            "max": 5000,
+            "strand": 1,
+            "attributes": {
+              "testid": ["t012", "t013", "t014"],
+              "gff_id": ["cds10001"],
+              "gff_name": ["edenprotein.1"]
+            }
+          }
+        },
+        "attributes": {
+          "testid": ["t004", "t001", "t004"],
+          "gff_id": ["mRNA10001"],
+          "gff_name": ["EDEN.1"]
+        }
+      }
+    },
+    "attributes": {
+      "testid": ["t003"],
+      "gff_id": ["gene10001"],
+      "gff_name": ["EDEN"]
+    }
+  }
+]

package/test_data/single_feature_no_children.gff3

@@ -0,0 +1 @@
+ctgA	example	remark	1000	2000	.	.	.	Name=Remark:hga;Alias=hga

package/test_data/single_feature_no_children.json

@@ -0,0 +1,14 @@
+[
+  {
+    "_id": "6931ef3bf5c33f70085c3a7b",
+    "refSeq": "ctgA",
+    "type": "remark",
+    "min": 999,
+    "max": 2000,
+    "attributes": {
+      "gff_source": ["example"],
+      "gff_name": ["Remark:hga"],
+      "gff_alias": ["hga"]
+    }
+  }
+]

package/test_data/single_feature_two_children.gff3

@@ -0,0 +1,3 @@
+ctgA	est	EST_match	1050	3202	.	+	.	ID=Match1;Name=agt830.5;Target=agt830.5 1 654
+ctgA	est	match_part	1050	1500	.	+	.	Parent=Match1;Name=agt830.5;Target=agt830.5 1 451
+ctgA	est	match_part	3000	3202	.	+	.	Parent=Match1;Name=agt830.5;Target=agt830.5 452 654

package/test_data/single_feature_two_children.json

@@ -0,0 +1,44 @@
+[
+  {
+    "_id": "6931fc4f88722ca882ba334c",
+    "refSeq": "ctgA",
+    "type": "EST_match",
+    "min": 1049,
+    "max": 3202,
+    "strand": 1,
+    "children": {
+      "6931fc4f88722ca882ba334a": {
+        "_id": "6931fc4f88722ca882ba334a",
+        "refSeq": "ctgA",
+        "type": "match_part",
+        "min": 1049,
+        "max": 1500,
+        "strand": 1,
+        "attributes": {
+          "gff_source": ["est"],
+          "gff_name": ["agt830.5"],
+          "gff_target": ["agt830.5 1 451"]
+        }
+      },
+      "6931fc4f88722ca882ba334b": {
+        "_id": "6931fc4f88722ca882ba334b",
+        "refSeq": "ctgA",
+        "type": "match_part",
+        "min": 2999,
+        "max": 3202,
+        "strand": 1,
+        "attributes": {
+          "gff_source": ["est"],
+          "gff_name": ["agt830.5"],
+          "gff_target": ["agt830.5 452 654"]
+        }
+      }
+    },
+    "attributes": {
+      "gff_source": ["est"],
+      "gff_id": ["Match1"],
+      "gff_name": ["agt830.5"],
+      "gff_target": ["agt830.5 1 654"]
+    }
+  }
+]

package/test_data/two_cds.gff3

@@ -1,9 +0,0 @@
-##gff-version 3
-##sequence-region   chr1 1000 9000
-#example01
-chr1	.	gene	1000	9000	.	+	.	ID=gene10001;Name=EDEN;testid=t003
-chr1	.	mRNA	1050	9000	.	+	.	ID=mRNA10001;Parent=gene10001;Name=EDEN.1;testid=t004,t001,t004
-chr1	.	exon	1050	1500	.	+	.	ID=exon10001;Parent=mRNA10001;testid=t007
-chr1	.	exon	5000	5500	.	+	.	ID=exon10004;Parent=mRNA10001;testid=t010
-chr1	.	CDS	1201	1500	.	+	0	ID=cds10001;Parent=mRNA10001;Name=edenprotein.1;testid=t012,t013,t014
-chr1	.	CDS	5000	5000	.	+	0	ID=cds10001;Parent=mRNA10001;Name=edenprotein.1;testid=t014

package/test_data/two_cds.json

@@ -1,67 +0,0 @@
-{
-  "_id": "66d70f3b9c7a7460925687a3",
-  "refSeq": "chr1",
-  "type": "gene",
-  "min": 999,
-  "max": 9000,
-  "strand": 1,
-  "children": {
-    "66d70f3b9c7a7460925687a2": {
-      "_id": "66d70f3b9c7a7460925687a2",
-      "refSeq": "chr1",
-      "type": "mRNA",
-      "min": 1049,
-      "max": 9000,
-      "strand": 1,
-      "children": {
-        "66d70f3b9c7a74609256879f": {
-          "_id": "66d70f3b9c7a74609256879f",
-          "refSeq": "chr1",
-          "type": "exon",
-          "min": 1049,
-          "max": 1500,
-          "strand": 1,
-          "attributes": {
-            "gff_id": ["exon10001"],
-            "testid": ["t007"]
-          }
-        },
-        "66d70f3b9c7a7460925687a0": {
-          "_id": "66d70f3b9c7a7460925687a0",
-          "refSeq": "chr1",
-          "type": "exon",
-          "min": 4999,
-          "max": 5500,
-          "strand": 1,
-          "attributes": {
-            "gff_id": ["exon10004"],
-            "testid": ["t010"]
-          }
-        },
-        "66d70f3b9c7a7460925687a1": {
-          "_id": "66d70f3b9c7a7460925687a1",
-          "refSeq": "chr1",
-          "type": "CDS",
-          "min": 1200,
-          "max": 5000,
-          "strand": 1,
-          "attributes": {
-            "gff_id": ["cds10001"],
-            "gff_name": ["edenprotein.1"],
-            "testid": ["t012", "t013", "t014"]
-          }
-        }
-      },
-      "attributes": {
-        "gff_id": ["mRNA10001"],
-        "gff_name": ["EDEN.1"],
-        "testid": ["t004", "t001", "t004"]
-      }
-    }
-  },
-  "attributes": {
-    "gff_id": ["gene10001"],
-    "gff_name": ["EDEN"],
-    "testid": ["t003"]
-  }
-}