MultiomicsClassifierQcChecker

URS & FS — спецификация пользователя и функциональная спецификация

Документ описывает контролируемый интерфейс и поведение утилиты MultiomicsClassifierQcChecker для сценария Multiomics Classifier QC Checker.

Предметные ограничения и критические параметры

• • Genomics Quality Score ≥ 0.80: Качество геномных данных (coverage, variant quality).
• • Transcriptomics Quality Score ≥ 0.80: Качество RNA-seq (RIN, mapping rate).
• • Methylomics Quality Score ≥ 0.80: Качество метилирования (conversion rate, CpG coverage).
• • Proteomics Quality Score ≥ 0.75: Качество протеомики (peptide IDs, FDR).
• • Fragmentomics Quality Score ≥ 0.80: Качество фрагментомики (fragment profile, UMI support).
• • Cross-Omics Concordance ≥ 0.70: Интегральная мера согласия между всеми доступными слоями.
• • Prediction Confidence ≥ 0.75: Уверенность ансамблевого предсказания.
• • Model Uncertainty ≤ 0.20: Эпистемическая + алеаторная неопределённость.
• • Minimum Required Layers ≥ 3: Минимальное число слоёв для надежного предсказания.
• ⚠️ ВАЖНО:
• • Высокая неопределённость модели = предсказание ненадёжно независимо от confidence score.
• 2. Missing Data Strategy: Определите заранее, как модель обрабатывает отсутствующие слои.

URS — пользовательские требования

Контракт input.csv

SampleID,ClassifierName,ClassifierVersion,IntegrationMethod,Genomics_Available,Genomics_QualityScore,MinGenomics_QualityScore,Genomics_Contribution_Weight,Transcriptomics_Available,Transcriptomics_QualityScore,MinTranscriptomics_QualityScore,Transcriptomics_Contribution_Weight,Methylomics_Available,Methylomics_QualityScore,MinMethylomics_QualityScore,Methylomics_Contribution_Weight,Proteomics_Available,Proteomics_QualityScore,MinProteomics_QualityScore,Proteomics_Contribution_Weight,Fragmentomics_Available,Fragmentomics_QualityScore,MinFragmentomics_QualityScore,Fragmentomics_Contribution_Weight,CrossOmics_Concordance_Score,MinCrossOmics_Concordance,DiscordantLayers,FinalPrediction_Score,PredictionConfidence,MinPredictionConfidence,ModelUncertainty,MaxModelUncertainty,PositiveControl_Pass,NegativeControl_Pass,CrossLayer_Control_Pass,LayersAvailable_Count,MinRequiredLayers,AllLayersFromSameSample
MO-2026-001,OncoMulti_v2,v2.1,Late_Fusion,true,0.95,0.80,0.30,true,0.91,0.80,0.25,true,0.88,0.80,0.20,true,0.85,0.75,0.15,true,0.92,0.80,0.10,0.89,0.70,,0.94,0.92,0.75,0.08,0.20,true,true,true,5,3,true
MO-2026-002,OncoMulti_v2,v2.1,Ensemble,true,0.92,0.80,0.35,true,0.87,0.80,0.25,true,0.62,0.80,0.20,false,0,0.75,0,true,0.88,0.80,0.20,0.45,0.70,Genomics-Methylomics,0.68,0.55,0.75,0.35,0.20,true,true,false,4,3,true
MO-2026-003,OncoMulti_v2,v2.1,Early_Fusion,true,0.90,0.80,0.60,false,0,0.80,0,false,0,0.80,0,false,0,0.75,0,true,0.85,0.80,0.40,0.82,0.70,,0.72,0.60,0.75,0.28,0.20,true,true,true,2,3,true

Правила валидации входных данных

FS — функциональная спецификация

Пример output.json

{
  "utilityId": "multiomicsclassifierqcchecker",
  "utilityFolder": "MultiomicsClassifierQcChecker",
  "package": "LiquidBiopsy",
  "overallStatus": "PASS|WARNING|FAIL",
  "sourceFile": "input.csv",
  "processedAtUtc": "2026-06-10T00:00:00Z",
  "checks": [
    {
      "parameter": "SampleID",
      "value": "MO-2026-001",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-001"
    },
    {
      "parameter": "ClassifierName",
      "value": "OncoMulti_v2",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-002"
    },
    {
      "parameter": "ClassifierVersion",
      "value": "v2.1",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-003"
    },
    {
      "parameter": "IntegrationMethod",
      "value": "Late_Fusion",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-004"
    },
    {
      "parameter": "Genomics_Available",
      "value": "true",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-005"
    },
    {
      "parameter": "Genomics_QualityScore",
      "value": "0.95",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-006"
    },
    {
      "parameter": "MinGenomics_QualityScore",
      "value": "0.80",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-007"
    },
    {
      "parameter": "Genomics_Contribution_Weight",
      "value": "0.30",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-008"
    },
    {
      "parameter": "Transcriptomics_Available",
      "value": "true",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-009"
    },
    {
      "parameter": "Transcriptomics_QualityScore",
      "value": "0.91",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-010"
    },
    {
      "parameter": "MinTranscriptomics_QualityScore",
      "value": "0.80",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-011"
    },
    {
      "parameter": "Transcriptomics_Contribution_Weight",
      "value": "0.25",
      "status": "PASS|WARNING|FAIL",
      "message": "Deterministic rule-based check result",
      "ruleReference": "FS-RULE-012"
    }
  ],
  "criticalFindings": [],
  "warnings": [],
  "audit": {
    "inputHash": "sha256:<calculated at runtime>",
    "rulesVersion": "<utility executable version>",
    "documentation": "MultiomicsClassifierQcChecker.documentation.html"
  }
}

Матрица трассируемости

IQ/OQ/PQ тестовые сценарии

QA/QC и change control

• Не менять имена колонок без обновления валидатора, документации и тестового набора.
• Хранить input.csv, output.json, версию исполняемого файла и контрольную сумму.
• Перед продуктивным использованием провести IQ/OQ/PQ или эквивалентную CSV/CSA-проверку.
• Критические пределы должны быть сверены с утверждённой спецификацией, регистрационным досье и локальными SOP.
• Утилита предоставляет формализованный QC decision support, но окончательное решение выпуска остаётся за QA/QP и утверждёнными процедурами.