離子軌跡偵測與訊號校正於高解析液相層析串聯飛行時間質譜儀平台之代謝體學應用

New Approaches for Ion Trace Detection and Signal Calibration for High Resolution Liquid Chromatography/Time-of-Flight Mass Spectrometry-Based Metabolomics Data

王三源(San-Yuan Wang)

指導教授 ： 曾宇鳳

國立臺灣大學/電機資訊學院/資訊工程學研究所/博士(2014年)

https://doi.org/10.6342/NTU.2014.10559

摘要

代謝體是一個用來研究表現型與發現生物標誌的工具。因液相層析串聯飛行時間質譜儀具備高敏感度、高再現性以及分析複雜生物樣本的能力。近來，液相層析串聯飛行時間質譜儀已成為用來偵測與定量小分子的重要技術。然而在液相層析串聯飛行時間質譜儀實驗的離子化步驟將產生大量的與代謝物無關的離子，因此在複雜的生物樣本裡偵測出代謝物仍然是一個很大的挑戰。大規模的流行病學研究包含了大量的生物樣本，這些樣本需要分成多個批次進行代謝體分析。但大規模的實驗結果往往會受不同的實驗批次與順序影響。我們開發PITracker演算法用來準確地從生物樣本中偵測出代謝物之離子軌跡，並且設計Batch Normalizer 演算法來校正大規模的代謝體實驗資料。
PITracker主要藉由四個步驟來偵測代謝物離子並精確地萃取出代謝物離子層析圖。首先，PITracker 藉由分析由液相層析串聯飛行時間質譜儀所產出質荷比(m/z)來推估相對質量差的容忍值。第二個步驟將根據最頻繁出現之基本離子(base ion)質荷比來調教質荷比已降低相對質量差。第三個步驟將依據步驟一所推估之相對質量差之容忍值偵測分析物之離子層析圖供一波峰偵測演算法(TIPick)使用。最後將根據一已知存在於該生物樣本之分析物(使用者指定)之質荷比來校正由波峰偵測演算法所回報之波峰質荷比。PITracker 的效能藉由373個人類代謝物標準品以及12個添加兩組54個不同濃度管制藥品之尿液樣本實驗來評估。這373個人類代謝物標準品包含5個因儀器飽和導致分離峰產生的標準品之離子層析圖都正確地被PITracker萃取出來了。此外管制藥品偵測之recall、precision 與F-score分別為0.94、1.00,與0.97均優於centWave之表現，更顯現出PITracker之穩定。
Batch Normalizer則考慮訊號總和之回歸模型增進其校正效能並移除批次與順序效應。Batch Normalizer 藉由23個匯聚品質管制樣本校正228個血液樣本之液相層析串聯飛行時間質譜資料。我們藉由校正前後之匯聚品質管制樣本偵測到的波峰訊號之相對標準差、任兩個匯聚品質管制樣本之皮爾森關係系數以及匯聚品質管制樣本在主成分分析圖之分布來評估其校正效能。校正後，匯聚品質管制樣本具相對標準差小於15%之波峰由11個增加到917個。所有匯聚品質管制樣本緊密地聚合在主成分分析圖上，而且兩兩匯聚品質管制樣本之平均皮爾森關係系數由0.938增加至0.976。同時我們亦比較了其它7種常用的校正方法，Batch Normalizer 明顯地具備最好的校正效能。

關鍵字

代謝體學 ； 層析圖 ； 質譜校正 ； 波峰偵測 ； 順序效應 ； 批次效應 ； 品質控制 ； 校正

並列摘要

Metabolomics is a powerful tool for understanding phenotypes and discovering biomarkers. Liquid Chromatography/Time-of-Flight Mass Spectrometry (LC/TOF-MS) has recently become an important technique to detected and quantify a broad range of small molecules due to their wide dynamic range, sensitivity, reproducibility, and ability to analyze complex biological fluids. However, to detect the metabolites in the complex biological samples with LC/TOF-MS is still a major challenge. This limitation can largely be attributed to the large amount of ions not produced by metabolites generated in the electrospray ionization process in LC/TOF-MS experiments. Combinations of multiple batches or data sets in large cross-sectional epidemiology studies are frequently utilized in metabolomics, but various systematic biases can introduce both batch and injection order effects and often require proper calibrations prior to chemometric analyses. We present PITracker, a novel algorithm that accurately and sensitively detects the ions produced by metabolites contained in complex biological samples and a novel algorithm, Batch Normalizer, to calibrate large-scale metabolomics data.
PITracker comprises four major steps to detect ions produced by metabolites and to extract pure ion chromatograms precisely. In the first step, PITracker analyzes all the m/z values in a raw data to estimate the adaptive relative mass difference tolerance of the ions from the same analyte in each LC/TOF-MS profile. In the second step, PITracker uses the most often present m/z value in base ions to calibrate m/z values to reduce the relative mass differences. In the third step, the pure ion chromatograms are extracted according to the estimated relative mass tolerance. Then a peak detection algorithm, TIPick, will be performed to detect the chromatographic peaks from the chromatograms. In the fourth step, the m/z values of the peaks reported by TIPick are corrected according to a known (user-specified) metabolite in samples. The performance of PITracker was tested in two data sets containing 373 human metabolite standards and 12 urine samples spiked of 54 forensic drugs with different concentrations. The utility of pure ion chromatogram extraction of PITracker was tested in the 373 human metabolite standards including 5 standards considered to be the split peaks due to instrumental saturation. The pure ion chromatograms of the 373 human metabolite standards were by PITracker correctly. The recall, precision, and F-score for forensic drug detection were 0.94, 1.00, and 0.97, respectively. Compared to centWave, we can demonstrate the robustness of PITracker.
Batch Normalizer utilizes a regression model with consideration of the total abundance of each sample to improve its calibration performance, and it is able to remove both batch effect and injection order effects. Batch Normalizer was tested using LC/TOF-MS data of 228 plasma samples and 23 pooled quality control (QC) samples. We evaluated the performance of Batch Normalizer by examining the distribution of relative standard deviation (RSD) for all peaks detected in the pooled QC samples, the average Pearson correlation coefficients for all peaks between any two of QC samples, and the distribution of QC samples in the scores plot of a principal component analysis (PCA). After calibration by Batch Normalizer, the number of peaks in QC samples with RSD less than 15% increased from 11 to 914, all of the QC samples were closely clustered in PCA scores plot, and the average Pearson correlation coefficients for all peaks of QC samples increased from 0.938 to 0.976. This method was compared to 7 commonly used calibration methods. We discovered that using Batch Normalizer to calibrate LC/TOF-MS data produces the best calibration results.

並列關鍵字

Metabolomics ； Chromatograms ； Mass Calibration ； Peak Detection ； Injection Order Effects ； Batch Effects ； Quality Control ； Calibration

參考文獻

[1] R. Ramanathan, M. Jemal, S. Ramagiri, Y. Q. Xia, W. G. Humpreys, T. Olah, and W. A. Korfmacher, "It is time for a paradigm shift in drug discovery bioanalysis: from SRM to HRMS," J Mass Spectrom, vol. 46, pp. 595-601, Jun 2011.

[2] J. L. Campbell and J. C. Le Blanc, "Using high-resolution quadrupole TOF technology in DMPK analyses," Bioanalysis, vol. 4, pp. 487-500, Mar 2012.

[3] S. Ojanpera and I. Ojanpera, "Forensic Drug Screening by LC–MS Using Accurate Mass Measurement," LC GC EUROPE, vol. 18, pp. 607-614, 2005.

[4] H. K. Lee, C. S. Ho, Y. P. H. Iu, P. S. J. Lai, C. C. Shek, Y.-C. Lo, H. B. Klinke, and M. Wood, "Development of a broad toxicological screening technique for urine using ultra-performance liquid chromatography and time-of-flight mass spectrometry," Analytica Chimica Acta, vol. 649, pp. 80-90, 2009.

[5] F. Hernandez, J. V. Sancho, M. Ibanez, and S. Grimalt, "Investigation of pesticide metabolites in food and water by LC-TOF-MS," TrAC Trends in Analytical Chemistry, vol. 27, pp. 862-872, 2008.