Whitin John C, Rangan Srinivasa, Cohen Harvey J
Department of Pediatrics, Stanford University School of Medicine, 300 Pasteur Drive, Stanford, CA USA.
BMC Res Notes. 2013 Sep 8;6:358. doi: 10.1186/1756-0500-6-358.
Biomarker discovery datasets created using mass spectrum protein profiling of complex mixtures of proteins contain many peaks that represent the same protein with different charge states. Correlated variables such as these can confound the statistical analyses of proteomic data. Previously we developed an algorithm that clustered mass spectrum peaks that were biologically or technically correlated. Here we demonstrate an algorithm that clusters correlated technical aliases only.
In this paper, we propose a preprocessing algorithm that can be used for grouping technical aliases in mass spectrometry protein profiling data. The stringency of the variance allowed for clustering is customizable, thereby affecting the number of peaks that are clustered. Subsequent analysis of the clusters, instead of individual peaks, helps reduce difficulties associated with technically-correlated data, and can aid more efficient biomarker identification.
This software can be used to pre-process and thereby decrease the complexity of protein profiling proteomics data, thus simplifying the subsequent analysis of biomarkers by decreasing the number of tests. The software is also a practical tool for identifying which features to investigate further by purification, identification and confirmation.
使用蛋白质复杂混合物的质谱蛋白质谱分析创建的生物标志物发现数据集包含许多代表具有不同电荷状态的同一蛋白质的峰。诸如此类的相关变量可能会混淆蛋白质组学数据的统计分析。我们之前开发了一种算法，该算法对在生物学或技术上相关的质谱峰进行聚类。在此，我们展示一种仅对相关技术别名进行聚类的算法。
在本文中，我们提出一种预处理算法，该算法可用于对质谱蛋白质谱分析数据中的技术别名进行分组。允许聚类的方差的严格程度是可定制的，从而影响聚类的峰的数量。对聚类而非单个峰进行后续分析有助于减少与技术相关数据相关的困难，并且可以帮助更有效地鉴定生物标志物。
该软件可用于预处理，从而降低蛋白质谱蛋白质组学数据的复杂性，通过减少测试数量简化生物标志物的后续分析。该软件也是一种实用工具，用于确定通过纯化、鉴定和确认进一步研究哪些特征。
