In metabolomics data, like other -omics data, normalization is an important\npart of the data processing. The goal of normalization is to reduce the variation\nfrom non-biological sources (such as instrument batch effects), while\nmaintaining the biological variation. Many normalization techniques make\nadjustments to each sample. One common method is to adjust each sample\nby its Total Ion Current (TIC), i.e. for each feature in the sample, divide its\nintensity value by the total for the sample. Because many of the assumptions\nof these methods are dubious in metabolomics data sets, we compare these\nmethods to two methods that make adjustments separately for each metabolite,\nrather than for each sample. These two methods are the following: 1) for\neach metabolite, divide its value by the median level in bridge samples\n(BRDG); 2) for each metabolite divide its value by the median across the experimental\nsamples (MED). These methods were assessed by comparing the\ncorrelation of the normalized values to the values from targeted assays for a\nsubset of metabolites in a large human plasma data set. The BRDG and MED\nnormalization techniques greatly outperformed the other methods, which often\nperformed worse than performing no normalization at all.
Loading....