Xia Qiangwei, Wang Tiansong, Park Yoonsuk, Lamont Richard J, Hackett Murray
Department of Chemical Engineering, University of Washington, Seattle, WA 98195, USA.
Int J Mass Spectrom. 2007 Jan 1;259(1-3):105-116. doi: 10.1016/j.ijms.2006.08.004.
Differential analysis of whole cell proteomes by mass spectrometry has largely been applied using various forms of stable isotope labeling. While metabolic stable isotope labeling has been the method of choice, it is often not possible to apply such an approach. Four different label free ways of calculating expression ratios in a classic "two-state" experiment are compared: signal intensity at the peptide level, signal intensity at the protein level, spectral counting at the peptide level, and spectral counting at the protein level. The quantitative data were mined from a dataset of 1245 qualitatively identified proteins, about 56% of the protein encoding open reading frames from Porphyromonas gingivalis, a Gram-negative intracellular pathogen being studied under extracellular and intracellular conditions. Two different control populations were compared against P. gingivalis internalized within a model human target cell line. The q-value statistic, a measure of false discovery rate previously applied to transcription microarrays, was applied to proteomics data. For spectral counting, the most logically consistent estimate of random error came from applying the locally weighted scatter plot smoothing procedure (LOWESS) to the most extreme ratios generated from a control technical replicate, thus setting upper and lower bounds for the region of experimentally observed random error.
通过质谱对全细胞蛋白质组进行差异分析，很大程度上是利用各种形式的稳定同位素标记来实现的。虽然代谢稳定同位素标记一直是首选方法，但往往无法采用这种方法。本文比较了经典“双态”实验中四种不同的无标记计算表达比率的方法：肽水平的信号强度、蛋白质水平的信号强度、肽水平的谱图计数以及蛋白质水平的谱图计数。定量数据来自一个包含1245个定性鉴定蛋白质的数据集，约占牙龈卟啉单胞菌蛋白质编码开放阅读框的56%，牙龈卟啉单胞菌是一种革兰氏阴性胞内病原体，正在细胞外和细胞内条件下进行研究。将两个不同的对照群体与内化于模型人类靶细胞系中的牙龈卟啉单胞菌进行比较。q值统计量（一种先前应用于转录微阵列的错误发现率度量）被应用于蛋白质组学数据。对于谱图计数，最符合逻辑的随机误差估计来自于对对照技术重复产生的最极端比率应用局部加权散点图平滑程序（LOWESS），从而为实验观察到的随机误差区域设定上下限。
