Department of Electrical Engineering, ESAT/SCD Katholieke Universiteit Leuven, Kasteelpark Arenberg 10, bus 2446, 3001 Heverlee, Leuven, Belgium.
J Am Soc Mass Spectrom. 2011 Mar;22(3):499-507. doi: 10.1007/s13361-010-0056-x. Epub 2011 Jan 15.
To reduce the influence of the between-spectra variability on the results of peptide quantification, one can consider the (18)O-labeling approach. Ideally, with such labeling technique, a mass shift of 4 Da of the isotopic distributions of peptides from the labeled sample is induced, which allows one to distinguish the two samples and to quantify the relative abundance of the peptides. It is worth noting, however, that the presence of small quantities of (16)O and (17)O atoms during the labeling step can cause incomplete labeling. In practice, ignoring incomplete labeling may result in the biased estimation of the relative abundance of the peptide in the compared samples. A Markov model was developed to address this issue (Zhu, Valkenborg, Burzykowski. J. Proteome Res. 9, 2669-2677, 2010). The model assumed that the peak intensities were normally distributed with heteroscedasticity using a power-of-the-mean variance funtion. Such a dependence has been observed in practice. Alternatively, we formulate the model within the Bayesian framework. This opens the possibility to further extend the model by the inclusion of random effects that can be used to capture the biological/technical variability of the peptide abundance. The operational characteristics of the model were investigated by applications to real-life mass-spectrometry data sets and a simulation study.
为了降低谱间变异性对肽定量结果的影响，可以考虑使用（18）O 标记方法。理想情况下，通过这种标记技术，可以诱导肽的同位素分布的质量位移为 4 Da，这允许区分两个样品并定量肽的相对丰度。然而，值得注意的是，在标记步骤中存在少量的（16）O 和（17）O 原子可能导致不完全标记。在实践中，忽略不完全标记可能会导致对比较样品中肽的相对丰度的有偏差的估计。已经开发了一种马尔可夫模型来解决这个问题（Zhu、Valkenborg、Burzykowski。J. Proteome Res. 9, 2669-2677, 2010）。该模型假设峰强度呈正态分布，具有异方差性，使用均值幂方差函数。在实践中已经观察到这种依赖性。或者，我们在贝叶斯框架内构建模型。这为通过包含可以用于捕获肽丰度的生物学/技术变异性的随机效应来进一步扩展模型提供了可能性。通过对真实质谱数据集和模拟研究的应用，研究了模型的操作特性。
