Identifying potential biomarkers in LC‐MS data

Abstract

Abstract In this paper an application of the uninformative variable elimination–partial least squares (UVE‐PLS) method extended by the Monte Carlo approach for selection of possible biomarkers from the liquid chromatography coupled with mass spectrometry (LC‐MS) data is reported. The main challenge consists not in the chemometrics analysis of LC‐MS data, but in the data organization. However, as demonstrated in our study, the selected variables are similar regardless of the data organization strategy. The best results are obtained for the standard normal variate (SNV) transformed data. Copyright © 2007 John Wiley & Sons, Ltd.