Quality assessment of LC-MS metabolomic data

Liquid Chromatography - Mass Spectrometry (LC-MS) is a promising high throughput technology to measure changes in metabolites present in human samples. However, LC-MS metabolomic data need to be carefully preprocessed to remove the effects of systematic bias and noise caused by the complex nature of human samples, presence of large number of metabolites, analytical variability and instrument noise. Without appropriate data preprocessing, these issues can mislead the interpretation of results from subsequent statistical analysis. In this study, we discuss problems with reproducibility of LC-MS metabolomic data and provide suggestions to detect and overcome these issues through quality assessment approaches and by means of appropriate experimental design. We present specific examples of the most common issues and discuss possible strategies to address them.