|Home | About | Journals | Submit | Contact Us | Français|
From the post-genomics and proteomics era, metabonomics/metabolomics has emerged as a vital area of research. Metabolic profiles of biological fluids contain a vast array of endogenous low-molecular-weight metabolites. Changes in these profiles resulting from perturbations of the system can be observed using information-rich analytical techniques, such as mass spectrometry. Due to the complexity of the samples, new separation techniques such as ultra-performance liquid chromatography have become accepted methods for these studies. A recent development, ion mobility mass spectrometry, is now also being employed to aid in extracting even more critical information from these sample sets.
The additional information obtained from these approaches has increased the complexity of the data, rendering them even more difficult to mine. Traditional profiling techniques, which involve scan-by-scan comparison of the data, have been used to compare small datasets; however, these approaches are not well suited to studies involving large numbers of samples with complex spectral information. Metabonomic/metabolomic approaches have been employed to mine large, complex datasets with great success. These approaches typically use multivariate statistical methods, such as principal component analysis (PCA), to highlight differences between samples based on observed spectral patterns. However, these methods are often not well suited to identifying subtle changes and can be biased by large variations within a sample class. New multivariate statistical methods, like orthogonal partial least squares, have been developed, which can overcome many of the problems observed when using PCA. We will illustrate these statistical and analytical methods with several examples obtained on a variety of sample types.