Many of the metabolomics methods are not directly quantitative. Therefore as many steps as possible in these methods should be controlled and internal and external controls should be used. Measurements performed by mass spectrometry based methods fluctuate in time and therefore quality control (QC) samples should regularly be analyzed during the run of related samples. Moreover, the samples should be randomized. The best randomization is not completely random but determined by the study design (number of samples per person, number of treatments) and the research question (are the differences within subject profiles important or explicitly the comparison between the subjects).

For the analytical design an important question that should be answered is whether you are interested in relative or absolute metabolite levels. If absolute metabolites levels are the goal, controls with known concentrations should be analyzed in the same runs.

In order to compare data from different runs and batches (even the run of this year with the one next year) an external reverence sample can be coanalyzed in every run. This may be in the form of a synthetically produced standard (Plasma standard metabolite mix) or a well stored large pool of reference sample (e.g. blood).