Journal of Proteome Research · 2012 · 50 citations · 41 references
Life SciencesBiochemistryOncologyGlycobiologyMass Spectrometric AnalysisNeoplasiaCancer GlycobiologyEsophageal CancerGlycan AnalysisGlycan StructureMass SpectrometryMass Spectrometry Data AnalysisGlycan FingerprintingIon MobilityIon Mobility SpectrometryMass Spectrometric Glycan AnalysisMass Spectrometric GlycomicsN-glycan ProfilingOutlier IdentificationOutlier Removal
Three disease phenotypes, Barrett's esophagus (BE), high-grade dysplasia (HGD), esophageal adenocarcinoma (EAC), and a set of normal control (NC) serum samples are examined using a combination of ion mobility spectrometry (IMS), mass spectrometry (MS), and principal component analysis (PCA) techniques. Samples from a total of 136 individuals were examined, including 7 characterized as BE, 12 as HGD, 56 as EAC, and 61 as NC. In typical data sets, it was possible to assign ∼20 to 30 glycan ions based on MS measurements. Ion mobility distributions for these ions show multiple features. In some cases, such as the [S1H5N4+3Na]3+ and [S1F1H5N4+3Na]3+ glycan ions, the ratio of intensities of high-mobility features to low-mobility features vary significantly for different groups. The degree to which such variations in mobility profiles can be used to distinguish phenotypes is evaluated for 11 N-linked glycan ions. An outlier analysis on each sample class followed by an unsupervised PCA using a genetic algorithm for pattern recognition reveals that EAC samples are separated from NC samples based on 46 features originating from the 11-glycan composite IMS distribution.
41
Colin Goodall, Ian T. Jolliffe · Technometrics · 1988 · 8.7K citations
Robust Regression and Outlier Detection
Gregory F. Piepel, Peter J. Rousseeuw, Annick M. Leroy · Technometrics · 1989 · 6.2K citations
G. Ya. Wiederschain · Biochemistry (Moscow) · 2009 · 4.1K citations · Full text