Interpretation framework

How to interpret Nanopore pathogen detection results

Note

Public This page provides a general framework for interpreting Nanopore sequencing results in pathogen detection, independent of specific experimental setups.

Overview

Nanopore sequencing 可以快速產生大量 reads,但:

detect ≠ true presence ≠ biological relevance

因此,結果判讀需要整合多個層面的證據,而不是依賴單一指標。


Core principle

Nanopore pathogen detection interpretation 應基於三個核心面向:

  1. Signal strength(訊號強度)
  2. Specificity(特異性)
  3. Context(生物與實驗脈絡)

👉 三者缺一不可


1️⃣ Signal strength

評估「訊號是否足夠支持存在」

常用指標:

  • read count(reads 數)
  • total bases
  • relative abundance
  • depth(如果有 reference)

Interpretation concept

  • 高 read count → 較可信
  • 極低 read count → 需謹慎(可能 background)
Warning

低 read count 的 detection 不應直接解讀為陽性。


2️⃣ Specificity

評估「訊號是否真的來自目標」

需要確認:

  • alignment quality(minimap2)
  • coverage pattern(是否均勻)
  • 是否為 conserved region(可能 cross-match)
  • database bias(Kraken2)

常見問題

  • 近緣物種誤判
  • conserved gene(如 rRNA)造成假陽性
  • k-mer overlap

👉 因此:

Kraken2 結果應由 alignment 驗證


3️⃣ Context

這是最重要、也最常被忽略的一層

需要考慮:

  • sample type(environment / tissue / culture)
  • expected microbiome
  • known contamination sources
  • negative control
  • experimental design

Example

  • 在 environmental sample 中偵測到低量 bacteria → 可能正常
  • 在 sterile tissue 中偵測到同樣 bacteria → 可能重要

Integration logic

實務上應整合三個面向:

Signal Specificity Context Interpretation
High High Consistent Strong evidence
Low High Consistent Possible presence
Low Low Inconsistent Likely noise
High Low Inconsistent Suspect artifact

Minimal interpretation workflow

flowchart TD
    A[Kraken2 detection] --> B[minimap2 validation]
    B --> C{Good alignment?}

    C -->|No| D[Likely false positive]
    C -->|Yes| E[Check read count]

    E --> F{Sufficient signal?}
    F -->|No| G[Low confidence]
    F -->|Yes| H[Check context]

    H --> I{Biologically plausible?}
    I -->|No| J[Possible contamination]
    I -->|Yes| K[Supported detection]


Common pitfalls

Warning

錯誤判讀通常來自「單一指標導向」。

  • 只看 Kraken2
  • 只看 read count
  • 不做 alignment
  • 忽略 negative control
  • 忽略 sample context

Practical guidelines(general)

以下為通用建議(非固定標準):

  • 不依賴單一 read 作為證據
  • 優先確認 alignment quality
  • 檢查 coverage 是否合理
  • 與 negative control 比較
  • 將結果放入 biological context 解讀

What this framework does NOT define

Note

This framework does not define strict thresholds.

以下項目需依不同實驗設定調整:

  • read count threshold
  • coverage cutoff
  • abundance cutoff
  • clinical significance criteria

👉 這些通常屬於:

  • specific project
  • clinical validation
  • regulatory setting

Quick takeaway

Tip

Nanopore 結果判讀不是單一數值判斷,而是多層證據整合。
Signal、specificity 與 context 必須同時考量,才能避免 false positive 或 over-interpretation。