Time-Saving Design of Experiment Protocol for Optimization of LC-MS Data Processing in Metabolomic Approaches
Time-Saving Design of Experiment Protocol for Optimization of LC-MS Data Processing in Metabolomic Approaches
复制标题
DOI：
10.1021/ac4020325
复制
发表时间：
2013-08-06
影响因子：
7.4
通讯作者：
Bertram, Hanne Christine
中科院分区：
文献类型：
--
作者：
Zheng, Hong;Clausen, Morten Rahr;Bertram, Hanne Christine
我们描述了一种省时的基于LC-MS的代谢组学数据处理方案，通过优化XCMS中的参数设置和使用实验设计(DoE)方法去除噪声和低强度峰的阈值设置，包括用于筛选的Plackett-Burman设计(PBD)和用于优化的中心复合设计(CCD)。基于对稀释序列的线性响应的评估的可靠性指数被用作评估数据质量的参数。在用PBD确定XCMS软件中的重要参数后，应用CCD法通过最大化信度和分组指数来确定它们的值。能源部的优化设置使标准混合物和人体尿液的可靠性指标分别比默认设置提高了19.4%和54.7%，总共需要38h才能完成优化。此外，利用电荷耦合器件对阈值设置进行了优化，进一步完善了阈值设置。将最优参数设置和阈值方法相结合的方法使标准混合物的可靠性指标提高了9.5倍，人体尿液数据的可靠性指标提高了14.5倍，总共需要41h。验证结果还表明，即使是对不同受试者的尿样，可靠性指标也提高了5~7倍。结论：所提出的方法可以作为一种节省时间的方法来改进基于LC-MS的代谢组学数据的处理。
We describe a time-saving protocol for the processing of LC-MS-based metabolomics data by optimizing parameter settings in XCMS and threshold settings for removing noisy and low-intensity peaks using design of experiment (DoE) approaches including Plackett-Burman design (PBD) for screening and central composite design (CCD) for optimization. A reliability index, which is based on evaluation of the linear response to a dilution series, was used as a parameter for the assessment of data quality. After identifying the significant parameters in the XCMS software by PBD, CCD was applied to determine their values by maximizing the reliability and group indexes. Optimal settings by DoE resulted in improvements of 19.4% and 54.7% in the reliability index for a standard mixture and human urine, respectively, as compared with the default setting, and a total of 38 h was required to complete the optimization. Moreover, threshold settings were optimized by using CCD for further improvement. The approach combining optimal parameter setting and the threshold method improved the reliability index about 9.5 times for a standards mixture and 14.5 times for human urine data, which required a total of 41 h. Validation results also showed improvements in the reliability index of about 5-7 times even for urine samples from different subjects. It is concluded that the proposed methodology can be used as a time-saving approach for improving the processing of LC-MS-based metabolomics data.