Institute of Chemistry, University of Silesia in Katowice, 9 Szkolna Street, 40-006 Katowice, Poland.
Molecules. 2021 Jan 25;26(3):621. doi: 10.3390/molecules26030621.
This article discusses the possibility of exploratory data analysis of samples described by second-order chromatographic data affected by peak shifts. In particular, the potential of the kernel Gram matrix representation as an alternative to the necessary and time-consuming alignment step is evaluated. It was demonstrated through several simulation studies and comparisons that even small peak shifts can be a substantial source of data variance, and they can easily hamper the interpretation of chromatographic data. When peak shifts are small, their negative effect is far more destructive than the impact of relatively large levels of the Gaussian noise, heteroscedastic noise, and signal's baseline. The Gram principal component analysis approach has proven to be a well-suited tool for exploratory analysis of chromatographic signals collected using the diode-array detector in which sample-to-sample peak shifts were observed.
本文讨论了对受峰位移影响的二阶色谱数据样本进行探索性数据分析的可能性。特别评估了核 Gram 矩阵表示作为必要且耗时的对齐步骤的替代方法的潜力。通过几项模拟研究和比较表明，即使是小的峰位移也可能是数据方差的主要来源，并且它们很容易阻碍色谱数据的解释。当峰位移较小时，其负面影响远比高斯噪声、异方差噪声和信号基线的相对较大水平的影响更为严重。Gram 主成分分析方法已被证明是一种非常适合的工具，可用于对使用二极管阵列检测器收集的色谱信号进行探索性分析，在这些信号中观察到了样品间的峰位移。
