@apollo-annotation/shared 0.1.19 → 0.1.20

package/src/GFF3/gff3ToAnnotationFeature.ts

@@ -80,8 +80,9 @@ function getFeatureMinMax(gff3Feature: GFF3Feature): [number, number] {
 
 function convertFeatureAttributes(
   gff3Feature: GFF3Feature,
-): Record<string, string[]> | undefined {
-  const convertedAttributes: Record<string, string[]> = {}
+): Record<string, string[] | undefined> | undefined {
+  const convertedAttributes: Record<string, string[] | undefined> | undefined =
+    {}
   const scores = gff3Feature
     .map((f) => f.score)
     .filter((score) => score !== null)
@@ -106,22 +107,21 @@ function convertFeatureAttributes(
   if (sources.length > 0) {
     let [source] = sources
     if (sources.length > 1) {
-      const sourceSet = new Set(...sources)
+      const sourceSet = new Set(sources)
       source = [...sourceSet].join(',')
     }
     convertedAttributes.gff_source = [source]
   }
   if (attributesCollections.length > 0) {
-    const newAttributes: Record<string, string[] | undefined> = {}
     for (const attributesCollection of attributesCollections) {
       for (const [key, val] of Object.entries(attributesCollection)) {
         if (!val || key === 'Parent') {
           continue
         }
         const newKey = isGFFReservedAttribute(key) ? gffToInternal[key] : key
-        const existingVal = newAttributes[newKey]
+        const existingVal = convertedAttributes[newKey]
         if (existingVal) {
-          const valSet = new Set(...existingVal, ...val)
+          const valSet = new Set([...existingVal, ...val])
           convertedAttributes[newKey] = [...valSet]
         } else {
           convertedAttributes[newKey] = val
@@ -158,7 +158,10 @@ function convertChildren(
     const [firstChildFeatureLocation] = childFeature
     if (
       firstChildFeatureLocation.type === 'three_prime_UTR' ||
-      firstChildFeatureLocation.type === 'five_prime_UTR'
+      firstChildFeatureLocation.type === 'five_prime_UTR' ||
+      firstChildFeatureLocation.type === 'intron' ||
+      firstChildFeatureLocation.type === 'start_codon' ||
+      firstChildFeatureLocation.type === 'stop_codon'
     ) {
       continue
     }
@@ -232,20 +235,17 @@ function processCDS(
       groupedLocations.push([location])
       continue
     }
-    const lastGroupLastLocation = lastGroup.at(-1)
-    if (!lastGroupLastLocation) {
-      throw new Error('Got group with no locations')
-    }
-    if (
+    const overlaps = lastGroup.some((lastGroupLoc) =>
       doesIntersect2(
         /* eslint-disable @typescript-eslint/no-non-null-assertion */
-        lastGroupLastLocation.start!,
-        lastGroupLastLocation.end!,
+        lastGroupLoc.start!,
+        lastGroupLoc.end!,
         location.start!,
         location.end!,
         /* eslint-enable @typescript-eslint/no-non-null-assertion */
-      )
-    ) {
+      ),
+    )
+    if (overlaps) {
       groupedLocations.push([location])
     } else {
       lastGroup.push(location)

package/src/util.ts

@@ -1,4 +1,5 @@
 /* eslint-disable @typescript-eslint/no-unsafe-assignment */
+
 import { AnnotationFeatureSnapshot } from '@apollo-annotation/mst'
 import { GFF3Feature } from '@gmod/gff'
 

package/test_data/braker.gff

@@ -0,0 +1,13 @@
+##gff-version	3
+CM033580.1	AUGUSTUS	gene	15529	16566	0.92	-	.	ID=g1;
+CM033580.1	AUGUSTUS	mRNA	15529	16566	0.92	-	.	ID=g1.t1;Parent=g1;
+CM033580.1	AUGUSTUS	stop_codon	15529	15531	.	-	0	ID=g1.t1.stop1;Parent=g1.t1;
+CM033580.1	AUGUSTUS	CDS	15529	15659	0.92	-	2	ID=g1.t1.CDS1;Parent=g1.t1;
+CM033580.1	AUGUSTUS	exon	15529	15659	.	-	.	ID=g1.t1.exon1;Parent=g1.t1;
+CM033580.1	AUGUSTUS	intron	15660	16112	0.96	-	.	ID=g1.t1.intron1;Parent=g1.t1;
+CM033580.1	AUGUSTUS	CDS	16113	16314	0.96	-	0	ID=g1.t1.CDS2;Parent=g1.t1;
+CM033580.1	AUGUSTUS	exon	16113	16314	.	-	.	ID=g1.t1.exon2;Parent=g1.t1;
+CM033580.1	AUGUSTUS	intron	16315	16536	0.96	-	.	ID=g1.t1.intron2;Parent=g1.t1;
+CM033580.1	AUGUSTUS	CDS	16537	16566	0.99	-	0	ID=g1.t1.CDS3;Parent=g1.t1;
+CM033580.1	AUGUSTUS	exon	16537	16566	.	-	.	ID=g1.t1.exon3;Parent=g1.t1;
+CM033580.1	AUGUSTUS	start_codon	16564	16566	.	-	0	ID=g1.t1.start1;Parent=g1.t1;

package/test_data/example01.gff3

@@ -0,0 +1,26 @@
+##gff-version 3
+##sequence-region   chr1 1000 9000
+#example01
+chr1	.	gene	1000	9000	.	+	.	ID=gene10001;Name=EDEN
+chr1	.	TF_binding_site	1000	1012	.	+	.	ID=tfbs10001;Parent=gene10001
+chr1	.	mRNA	1050	9000	.	+	.	ID=mRNA10001;Parent=gene10001;Name=EDEN.1
+chr1	.	mRNA	1050	9000	.	+	.	ID=mRNA10002;Parent=gene10001;Name=EDEN.2
+chr1	.	mRNA	1300	9000	.	+	.	ID=mRNA10003;Parent=gene10001;Name=EDEN.3
+chr1	.	exon	1050	1500	.	+	.	ID=exon10001;Parent=mRNA10001,mRNA10002
+chr1	.	exon	1300	1500	.	+	.	ID=exon10002;Parent=mRNA10003
+chr1	.	exon	3000	3902	.	+	.	ID=exon10003;Parent=mRNA10001,mRNA10003
+chr1	.	exon	5000	5500	.	+	.	ID=exon10004;Parent=mRNA10001,mRNA10002,mRNA10003
+chr1	.	exon	7000	9000	.	+	.	ID=exon10005;Parent=mRNA10001,mRNA10002,mRNA10003
+chr1	.	CDS	1201	1500	.	+	0	ID=cds10001;Parent=mRNA10001;Name=edenprotein.1
+chr1	.	CDS	3000	3902	.	+	0	ID=cds10001;Parent=mRNA10001;Name=edenprotein.1
+chr1	.	CDS	5000	5500	.	+	0	ID=cds10001;Parent=mRNA10001;Name=edenprotein.1
+chr1	.	CDS	7000	7600	.	+	0	ID=cds10001;Parent=mRNA10001;Name=edenprotein.1
+chr1	.	CDS	1201	1500	.	+	0	ID=cds10002;Parent=mRNA10002;Name=edenprotein.2
+chr1	.	CDS	5000	5500	.	+	0	ID=cds10002;Parent=mRNA10002;Name=edenprotein.2
+chr1	.	CDS	7000	7600	.	+	0	ID=cds10002;Parent=mRNA10002;Name=edenprotein.2
+chr1	.	CDS	3301	3902	.	+	0	ID=cds10003;Parent=mRNA10003;Name=edenprotein.3
+chr1	.	CDS	5000	5500	.	+	1	ID=cds10003;Parent=mRNA10003;Name=edenprotein.3
+chr1	.	CDS	7000	7600	.	+	1	ID=cds10003;Parent=mRNA10003;Name=edenprotein.3
+chr1	.	CDS	3391	3902	.	+	0	ID=cds10004;Parent=mRNA10003;Name=edenprotein.4
+chr1	.	CDS	5000	5500	.	+	1	ID=cds10004;Parent=mRNA10003;Name=edenprotein.4
+chr1	.	CDS	7000	7600	.	+	1	ID=cds10004;Parent=mRNA10003;Name=edenprotein.4

package/test_data/example01.json

@@ -0,0 +1,236 @@
+{
+  "_id": "66e049f17b9cedae9ad89108",
+  "refSeq": "chr1",
+  "type": "gene",
+  "min": 999,
+  "max": 9000,
+  "strand": 1,
+  "children": {
+    "66e049f17b9cedae9ad890f5": {
+      "_id": "66e049f17b9cedae9ad890f5",
+      "refSeq": "chr1",
+      "type": "TF_binding_site",
+      "min": 999,
+      "max": 1012,
+      "strand": 1,
+      "attributes": {
+        "gff_id": ["tfbs10001"]
+      }
+    },
+    "66e049f17b9cedae9ad890fb": {
+      "_id": "66e049f17b9cedae9ad890fb",
+      "refSeq": "chr1",
+      "type": "mRNA",
+      "min": 1049,
+      "max": 9000,
+      "strand": 1,
+      "children": {
+        "66e049f17b9cedae9ad890f6": {
+          "_id": "66e049f17b9cedae9ad890f6",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 1049,
+          "max": 1500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10002"]
+          }
+        },
+        "66e049f17b9cedae9ad890f7": {
+          "_id": "66e049f17b9cedae9ad890f7",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 2999,
+          "max": 3902,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10003"]
+          }
+        },
+        "66e049f17b9cedae9ad890f8": {
+          "_id": "66e049f17b9cedae9ad890f8",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 4999,
+          "max": 5500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10004"]
+          }
+        },
+        "66e049f17b9cedae9ad890f9": {
+          "_id": "66e049f17b9cedae9ad890f9",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 6999,
+          "max": 9000,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10005"]
+          }
+        },
+        "66e049f17b9cedae9ad890fa": {
+          "_id": "66e049f17b9cedae9ad890fa",
+          "refSeq": "chr1",
+          "type": "CDS",
+          "min": 1200,
+          "max": 7600,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["cds10001"],
+            "gff_name": ["edenprotein.1"]
+          }
+        }
+      },
+      "attributes": {
+        "gff_id": ["mRNA10001"],
+        "gff_name": ["EDEN.1"]
+      }
+    },
+    "66e049f17b9cedae9ad89100": {
+      "_id": "66e049f17b9cedae9ad89100",
+      "refSeq": "chr1",
+      "type": "mRNA",
+      "min": 1049,
+      "max": 9000,
+      "strand": 1,
+      "children": {
+        "66e049f17b9cedae9ad890fc": {
+          "_id": "66e049f17b9cedae9ad890fc",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 1049,
+          "max": 1500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10002"]
+          }
+        },
+        "66e049f17b9cedae9ad890fd": {
+          "_id": "66e049f17b9cedae9ad890fd",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 4999,
+          "max": 5500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10004"]
+          }
+        },
+        "66e049f17b9cedae9ad890fe": {
+          "_id": "66e049f17b9cedae9ad890fe",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 6999,
+          "max": 9000,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10005"]
+          }
+        },
+        "66e049f17b9cedae9ad890ff": {
+          "_id": "66e049f17b9cedae9ad890ff",
+          "refSeq": "chr1",
+          "type": "CDS",
+          "min": 1200,
+          "max": 7600,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["cds10002"],
+            "gff_name": ["edenprotein.2"]
+          }
+        }
+      },
+      "attributes": {
+        "gff_id": ["mRNA10002"],
+        "gff_name": ["EDEN.2"]
+      }
+    },
+    "66e049f17b9cedae9ad89107": {
+      "_id": "66e049f17b9cedae9ad89107",
+      "refSeq": "chr1",
+      "type": "mRNA",
+      "min": 1299,
+      "max": 9000,
+      "strand": 1,
+      "children": {
+        "66e049f17b9cedae9ad89101": {
+          "_id": "66e049f17b9cedae9ad89101",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 1299,
+          "max": 1500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10001"]
+          }
+        },
+        "66e049f17b9cedae9ad89102": {
+          "_id": "66e049f17b9cedae9ad89102",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 2999,
+          "max": 3902,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10003"]
+          }
+        },
+        "66e049f17b9cedae9ad89103": {
+          "_id": "66e049f17b9cedae9ad89103",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 4999,
+          "max": 5500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10004"]
+          }
+        },
+        "66e049f17b9cedae9ad89104": {
+          "_id": "66e049f17b9cedae9ad89104",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 6999,
+          "max": 9000,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon10005"]
+          }
+        },
+        "66e049f17b9cedae9ad89105": {
+          "_id": "66e049f17b9cedae9ad89105",
+          "refSeq": "chr1",
+          "type": "CDS",
+          "min": 3300,
+          "max": 7600,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["cds10003"],
+            "gff_name": ["edenprotein.3"]
+          }
+        },
+        "66e049f17b9cedae9ad89106": {
+          "_id": "66e049f17b9cedae9ad89106",
+          "refSeq": "chr1",
+          "type": "CDS",
+          "min": 3390,
+          "max": 7600,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["cds10004"],
+            "gff_name": ["edenprotein.4"]
+          }
+        }
+      },
+      "attributes": {
+        "gff_id": ["mRNA10003"],
+        "gff_name": ["EDEN.3"]
+      }
+    }
+  },
+  "attributes": {
+    "gff_id": ["gene10001"],
+    "gff_name": ["EDEN"]
+  }
+}

package/test_data/example02.json

@@ -0,0 +1,225 @@
+{
+  "_id": "66e049609048deab4117a33e",
+  "refSeq": "chr1",
+  "type": "gene",
+  "min": 10999,
+  "max": 19000,
+  "strand": 1,
+  "children": {
+    "66e049609048deab4117a331": {
+      "_id": "66e049609048deab4117a331",
+      "refSeq": "chr1",
+      "type": "mRNA",
+      "min": 11049,
+      "max": 19000,
+      "strand": 1,
+      "children": {
+        "66e049609048deab4117a32c": {
+          "_id": "66e049609048deab4117a32c",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 11049,
+          "max": 11500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20001"]
+          }
+        },
+        "66e049609048deab4117a32d": {
+          "_id": "66e049609048deab4117a32d",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 12999,
+          "max": 13902,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20004"]
+          }
+        },
+        "66e049609048deab4117a32e": {
+          "_id": "66e049609048deab4117a32e",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 14999,
+          "max": 15500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20006"]
+          }
+        },
+        "66e049609048deab4117a32f": {
+          "_id": "66e049609048deab4117a32f",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 16999,
+          "max": 19000,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20009"]
+          }
+        },
+        "66e049609048deab4117a330": {
+          "_id": "66e049609048deab4117a330",
+          "refSeq": "chr1",
+          "type": "CDS",
+          "min": 11200,
+          "max": 17600,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["cds20001"],
+            "gff_name": ["edenprotein.1"]
+          }
+        }
+      },
+      "attributes": {
+        "gff_id": ["mRNA20001"],
+        "gff_name": ["EDEN.1"]
+      }
+    },
+    "66e049609048deab4117a336": {
+      "_id": "66e049609048deab4117a336",
+      "refSeq": "chr1",
+      "type": "mRNA",
+      "min": 11049,
+      "max": 19000,
+      "strand": 1,
+      "children": {
+        "66e049609048deab4117a332": {
+          "_id": "66e049609048deab4117a332",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 11049,
+          "max": 11500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20002"]
+          }
+        },
+        "66e049609048deab4117a333": {
+          "_id": "66e049609048deab4117a333",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 14999,
+          "max": 15500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20007"]
+          }
+        },
+        "66e049609048deab4117a334": {
+          "_id": "66e049609048deab4117a334",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 16999,
+          "max": 19000,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20010"]
+          }
+        },
+        "66e049609048deab4117a335": {
+          "_id": "66e049609048deab4117a335",
+          "refSeq": "chr1",
+          "type": "CDS",
+          "min": 11200,
+          "max": 17600,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["cds20002"],
+            "gff_name": ["edenprotein.2"]
+          }
+        }
+      },
+      "attributes": {
+        "gff_id": ["mRNA20002"],
+        "gff_name": ["EDEN.2"]
+      }
+    },
+    "66e049609048deab4117a33d": {
+      "_id": "66e049609048deab4117a33d",
+      "refSeq": "chr1",
+      "type": "mRNA",
+      "min": 11299,
+      "max": 19000,
+      "strand": 1,
+      "children": {
+        "66e049609048deab4117a337": {
+          "_id": "66e049609048deab4117a337",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 11299,
+          "max": 11500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20003"]
+          }
+        },
+        "66e049609048deab4117a338": {
+          "_id": "66e049609048deab4117a338",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 12999,
+          "max": 13902,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20005"]
+          }
+        },
+        "66e049609048deab4117a339": {
+          "_id": "66e049609048deab4117a339",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 14999,
+          "max": 15500,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20008"]
+          }
+        },
+        "66e049609048deab4117a33a": {
+          "_id": "66e049609048deab4117a33a",
+          "refSeq": "chr1",
+          "type": "exon",
+          "min": 16999,
+          "max": 19000,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["exon20011"]
+          }
+        },
+        "66e049609048deab4117a33b": {
+          "_id": "66e049609048deab4117a33b",
+          "refSeq": "chr1",
+          "type": "CDS",
+          "min": 13300,
+          "max": 17600,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["cds20003"],
+            "gff_name": ["edenprotein.3"]
+          }
+        },
+        "66e049609048deab4117a33c": {
+          "_id": "66e049609048deab4117a33c",
+          "refSeq": "chr1",
+          "type": "CDS",
+          "min": 13390,
+          "max": 17600,
+          "strand": 1,
+          "attributes": {
+            "gff_id": ["cds20004"],
+            "gff_name": ["edenprotein.4"]
+          }
+        }
+      },
+      "attributes": {
+        "gff_id": ["mRNA20003"],
+        "gff_name": ["EDEN.3"]
+      }
+    }
+  },
+  "attributes": {
+    "gff_id": ["gene20001"],
+    "gff_name": ["EDEN"]
+  }
+}