Merge pull request #27 from molgenis/fix/format_multi_missing

Fix #26 missing values for format fields with multiple values
molgenis · Jul 6, 2023 · 3ad10b2 · 3ad10b2
2 parents 408fb0d + f8cbb14
commit 3ad10b2
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 3 deletions.
diff --git a/src/VcfWriter.ts b/src/VcfWriter.ts
@@ -144,14 +144,19 @@ function writeFieldValueSingle(field: FieldMetadata, value: Value, missingValue
   return vcf;
 }
 
-function writeFieldValueMultiple(field: FieldMetadata, values: ValueArray, separator: string): string {
+function writeFieldValueMultiple(
+  field: FieldMetadata,
+  values: ValueArray,
+  separator: string,
+  missingValue = MISSING
+): string {
   const vcf = [];
 
   for (const infoValue of values) {
     if (field.nested) {
       vcf.push(writeFieldValueNested(field.nested, infoValue as ValueArray));
     } else {
-      vcf.push(writeFieldValue(field, infoValue, ""));
+      vcf.push(writeFieldValue(field, infoValue, missingValue));
     }
   }
 
@@ -164,7 +169,7 @@ function writeFieldValueNested(nestedField: NestedFieldMetadata, nestedValues: V
     if (infoField.number.count === 1) {
       vcf.push(writeFieldValueSingle(infoField, nestedValues[index], ""));
     } else {
-      vcf.push(writeFieldValueMultiple(infoField, nestedValues[index] as ValueArray, "&"));
+      vcf.push(writeFieldValueMultiple(infoField, nestedValues[index] as ValueArray, "&", ""));
     }
   }
   return vcf.join(nestedField.separator);

diff --git a/src/__tests__/VcfWriter.test.ts b/src/__tests__/VcfWriter.test.ts
@@ -56,6 +56,18 @@ test("parse and write vcf: Samples none", () => {
   expect(writeVcf(parseVcf(vcfSamples), { samples: [] })).toBe(expectedVcfSamples);
 });
 
+test("parse and write vcf: Samples missing values", () => {
+  const expectedVcfSamplesMissingValues = `##fileformat=VCFv4.2
+##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
+##FORMAT=<ID=HQ,Number=2,Type=Integer,Description="Haplotype Quality">
+##FORMAT=<ID=GQ,Number=1,Type=Integer,Description="Genotype Quality">
+##contig=<ID=1,length=249250621,assembly=b37>
+#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tSAMPLE0\tSAMPLE1\tSAMPLE2
+1\t1\t.\tA\tG\t.\t.\t.\tGT:HQ:GQ\t0|1:.:1\t0/1:.,.:2\t1/1:.,4:3
+`;
+  expect(writeVcf(parseVcf(vcfSamplesMissingValues))).toBe(expectedVcfSamplesMissingValues);
+});
+
 const vcfIdRefAltQualFilter = `##fileformat=VCFv4.2
 ##FILTER=<ID=q10,Description="Quality below 10">
 ##FILTER=<ID=q20,Description="Quality below 20">
@@ -131,3 +143,12 @@ const vcfSamples = `##fileformat=VCFv4.2
 #CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tSAMPLE0\tSAMPLE1\tSAMPLE2
 1\t1\t.\tA\tG\t.\t.\t.\tGT:AD:GQ:HQ\t0|1:1,50:1:2,3\t0/1:20,80:.:4,5\t1/1
 `;
+
+const vcfSamplesMissingValues = `##fileformat=VCFv4.2
+##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
+##FORMAT=<ID=HQ,Number=2,Type=Integer,Description="Haplotype Quality">
+##FORMAT=<ID=GQ,Number=1,Type=Integer,Description="Genotype Quality">
+##contig=<ID=1,length=249250621,assembly=b37>
+#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tSAMPLE0\tSAMPLE1\tSAMPLE2
+1\t1\t.\tA\tG\t.\t.\t.\tGT:HQ:GQ\t0|1:.:1\t0/1:.,.:2\t1/1:.,4:3
+`;