Evaluation of normalization methods for analysis of LC-MS data

The purpose of normalization of data generated by liquid chromatography coupled with mass spectrometry (LC-MS) is to reduce bias due to differences in sample collection, biomolecule extraction, and instrument variability. In this paper several normalization methods are reviewed and evaluated based on LC-MS data acquired from experimental and quality control (QC) samples. Specifically, LC-MS data from a metabolomic study aimed at discovering liver cancer biomarkers are analyzed to evaluate the performance of the normalization methods. ANOVA models are used for identification of ions with statistically significant peak intensities between liver cancer and cirrhotic controls. Also, LC-MS data from QC samples are analyzed to assess the ability of the normalization methods in decreasing the variability of ion intensity measurements in multiple runs. Significant run to run variability is observed despite normalizing the LC-MS data by various methods. Thus, it is important to select a suitable normalization method for each data set, as it is difficult to find a method that is applicable for all types of LC-MS data.