feat: tweak parameters of QC rules. remove unused config values

nextstrain · Aug 21, 2020 · b78b143 · b78b143
1 parent 8d49eee
commit b78b143
Show file tree

Hide file tree

Showing 5 changed files with 13 additions and 22 deletions.
diff --git a/packages/web/src/algorithms/QC/ruleMixedSites.ts b/packages/web/src/algorithms/QC/ruleMixedSites.ts
@@ -4,15 +4,13 @@ import type { AnalysisResult, NucleotideSubstitution } from 'src/algorithms/type
 
 export interface QCRulesConfigMixedSites {
   mixedSitesThreshold: number
-  scoreWeight: number
-  scoreBias: number
   scoreMax: number
 }
 
 export function ruleMixedSites(
   { nucleotideComposition }: AnalysisResult,
   _1: NucleotideSubstitution[],
-  { mixedSitesThreshold, scoreWeight, scoreBias, scoreMax }: QCRulesConfigMixedSites,
+  { mixedSitesThreshold, scoreMax }: QCRulesConfigMixedSites,
 ) {
   const goodBases = new Set(['A', 'C', 'G', 'T', 'N', '-'])
 
@@ -21,9 +19,8 @@ export function ruleMixedSites(
     .reduce((a, b) => a + nucleotideComposition[b], 0)
 
   let scoreRaw = 0
-  if (totalMixedSites > mixedSitesThreshold) {
-    scoreRaw = (totalMixedSites - mixedSitesThreshold) * scoreWeight - scoreBias
-  }
+  scoreRaw = 100 * (totalMixedSites / mixedSitesThreshold)
+
   const score = clamp(scoreRaw, 0, scoreMax)
 
   return { score, totalMixedSites, mixedSitesThreshold }

diff --git a/packages/web/src/algorithms/QC/ruleSnpClusters.ts b/packages/web/src/algorithms/QC/ruleSnpClusters.ts
@@ -51,9 +51,7 @@ export function processSNPClusters(snpClusters: number[][]): ClusteredSNPs[] {
 export interface QCRulesConfigSNPClusters {
   windowSize: number
   clusterCutOff: number
-  totalSNPsThreshold: number
   scoreWeight: number
-  scoreBias: number
   scoreMax: number
 }
 
@@ -62,19 +60,18 @@ export function ruleSnpClusters(
   privateMutations: NucleotideSubstitution[],
   config: QCRulesConfigSNPClusters,
 ) {
-  const { totalSNPsThreshold, scoreWeight, scoreBias, scoreMax } = config
+  const { scoreWeight, scoreMax } = config
 
   const snpClusters = findSNPClusters(data, privateMutations, config)
   const clusteredSNPs = processSNPClusters(snpClusters)
   const totalSNPs = clusteredSNPs.reduce((acc, { numberOfSNPs }) => acc + numberOfSNPs, 0)
 
   let scoreRaw = 0
-  if (totalSNPs > totalSNPsThreshold) {
-    scoreRaw = (totalSNPs - totalSNPsThreshold) * scoreWeight - scoreBias
-  }
+  scoreRaw = snpClusters.length * scoreWeight
+
   const score = clamp(scoreRaw, 0, scoreMax)
 
-  return { score, totalSNPs, totalSNPsThreshold, clusteredSNPs }
+  return { score, totalSNPs, clusteredSNPs }
 }
 
 export type QCResultSNPClusters = ReturnType<typeof ruleSnpClusters>
diff --git a/packages/web/src/algorithms/QC/runQC.ts b/packages/web/src/algorithms/QC/runQC.ts
@@ -37,18 +37,14 @@ const qcRulesConfigDefault: QCRulesConfig = {
   },
   snpClusters: {
     enabled: true,
-    totalSNPsThreshold: 0,
     windowSize: 100, // window along the genome to look for a cluster
     clusterCutOff: 4, // number of mutations within that window to trigger a cluster
-    scoreWeight: 1,
-    scoreBias: 0,
+    scoreWeight: 50, // each cluster counts for 50
     scoreMax: Infinity,
   },
   mixedSites: {
     enabled: true,
     mixedSitesThreshold: 10, // number of non-ACGTN sites to trigger warning
-    scoreWeight: 1,
-    scoreBias: 0,
     scoreMax: Infinity,
   },
 } as const

diff --git a/packages/web/src/helpers/formatQCMixedSites.ts b/packages/web/src/helpers/formatQCMixedSites.ts
@@ -14,7 +14,8 @@ export function formatQCMixedSites<TFunction extends TFunctionInterface>(
   }
 
   const { score, totalMixedSites, mixedSitesThreshold } = mixedSites
-  return t('Too many mixed sites: Total mixed: {{total}} ({{allowed}} allowed). QC score: {{score}}', {
+  return t('{{warn}} mixed sites: Total mixed: {{total}} ({{allowed}} allowed). QC score: {{score}}', {
+    warn: score > 100 ? 'Too many' : 'Found',
     total: totalMixedSites,
     allowed: mixedSitesThreshold,
     score: round(score),

diff --git a/packages/web/src/helpers/formatQCSNPClusters.ts b/packages/web/src/helpers/formatQCSNPClusters.ts
@@ -13,10 +13,10 @@ export function formatQCSNPClusters<TFunction extends TFunctionInterface>(
     return undefined
   }
 
-  const { score, totalSNPs, totalSNPsThreshold } = snpClusters
-  return t('Too many SNP clusters. Total clusters: {{total}} ({{allowed}} allowed). QC score: {{score}}', {
+  const { score, totalSNPs } = snpClusters
+  return t('Found {{nClusters}} SNP clusters with total of {{total}} mutations. QC score: {{score}}', {
     total: totalSNPs,
-    allowed: totalSNPsThreshold,
+    nClusters: snpClusters.clusteredSNPs.length,
     score: round(score),
   })
 }