Better handling of cases of variants having missing FORMAT fields (#113)

Author: cmdcolin

package.json

@@ -55,7 +55,7 @@
     "@vitest/coverage-v8": "^3.0.4",
     "documentation": "^14.0.1",
     "eslint": "^9.7.0",
-    "eslint-plugin-unicorn": "^57.0.0",
+    "eslint-plugin-unicorn": "^58.0.0",
     "prettier": "^3.2.4",
     "rimraf": "^6.0.1",
     "standard-changelog": "^6.0.0",

src/parse.ts

@@ -1,3 +1,4 @@
+import { parseGenotypesOnly } from './parseGenotypesOnly.ts'
 import { parseMetaString } from './parseMetaString.ts'
 import vcfReserved from './vcfReserved.ts'
 
@@ -120,38 +121,6 @@ export default class VCFParser {
     return genotypes
   }
 
-  private parseGenotypesOnly(format: string, prerest: string) {
-    const rest = prerest.split('\t')
-    const genotypes = {} as Record<string, string>
-    let i = 0
-    const formatSplit = format.split(':')
-    if (formatSplit.length === 1) {
-      for (const sample of this.samples) {
-        genotypes[sample] = rest[i++]!
-      }
-    } else {
-      const gtIndex = formatSplit.indexOf('GT')
-      if (gtIndex === 0) {
-        for (const sample of this.samples) {
-          const val = rest[i++]!
-          const idx = val.indexOf(':')
-          if (idx !== -1) {
-            genotypes[sample] = val.slice(0, idx)
-          } else {
-            console.warn('unknown')
-          }
-        }
-      } else {
-        for (const sample of this.samples) {
-          const val = rest[i++]!.split(':')
-          genotypes[sample] = val[gtIndex]!
-        }
-      }
-    }
-
-    return genotypes
-  }
-
   /**
    * Parse a VCF metadata line (i.e. a line that starts with "##") and add its
    * properties to the object.
@@ -338,7 +307,7 @@ export default class VCFParser {
       QUAL: qual,
       FORMAT: format,
       SAMPLES: () => this.parseSamples(fields[8] ?? '', rest),
-      GENOTYPES: () => this.parseGenotypesOnly(fields[8] ?? '', rest),
+      GENOTYPES: () => parseGenotypesOnly(fields[8] ?? '', rest, this.samples),
     }
   }
 }

src/parseGenotypesOnly.ts

@@ -0,0 +1,33 @@
+export function parseGenotypesOnly(
+  format: string,
+  prerest: string,
+  samples: string[],
+) {
+  const rest = prerest.split('\t')
+  const genotypes = {} as Record<string, string>
+  let i = 0
+  if (format.includes('GT')) {
+    const formatSplit = format.split(':')
+    if (formatSplit.length === 1) {
+      for (const sample of samples) {
+        genotypes[sample] = rest[i++]!
+      }
+    } else {
+      const gtIndex = formatSplit.indexOf('GT')
+      if (gtIndex === 0) {
+        for (const sample of samples) {
+          const val = rest[i++]!
+          const idx = val.indexOf(':')
+          genotypes[sample] = idx !== -1 ? val.slice(0, idx) : val
+        }
+      } else {
+        for (const sample of samples) {
+          const val = rest[i++]!.split(':')
+          genotypes[sample] = val[gtIndex]!
+        }
+      }
+    }
+  }
+
+  return genotypes
+}

test/__snapshots__/parseGenotypesOnly.test.ts.snap

@@ -0,0 +1,22 @@
+// Vitest Snapshot v1, https://vitest.dev/guide/snapshot.html
+
+exports[`parse genotypes 1`] = `
+{
+  "h1": "./.",
+  "h2": "./.",
+}
+`;
+
+exports[`parse genotypes 2`] = `
+{
+  "h1": "./.",
+  "h2": "./.",
+}
+`;
+
+exports[`parse genotypes 3`] = `
+{
+  "h1": "./.",
+  "h2": "./.",
+}
+`;

test/parseGenotypesOnly.test.ts

@@ -0,0 +1,12 @@
+import { test, expect } from 'vitest'
+import { parseGenotypesOnly } from '../src/parseGenotypesOnly'
+
+test('parse genotypes', () => {
+  expect(parseGenotypesOnly('GT', './.\t./.', ['h1', 'h2'])).toMatchSnapshot()
+  expect(
+    parseGenotypesOnly('GT:RT', './.:1\t./.', ['h1', 'h2']),
+  ).toMatchSnapshot()
+  expect(
+    parseGenotypesOnly('RT:GT', '1:./.\t2:./.', ['h1', 'h2']),
+  ).toMatchSnapshot()
+})

tsconfig.json

@@ -3,6 +3,7 @@
   "compilerOptions": {
     "outDir": "dist",
     "lib": ["dom", "esnext"],
+    "skipLibCheck": true,
     "declaration": true,
     "moduleResolution": "node",
     "sourceMap": true,