Instrumental and experimental effects in LC-MS-based metabolomics.

The experimental complexity of a metabolomics study can cause uncontrolled variance that is not related to the biological effect being studied and may distort or obscure the data analysis. While some sources can be controlled with good experimental techniques and careful sample handling, others are inherent in the analytical technique used and cannot easily be avoided. We discuss the sources and appearance of some of these artifacts and show ways in which they can be detected using visualization and statistical tools, allowing appropriate treatment prior to multivariate analysis (MVA).