Mean Center
Many classification methods are influenced by the magnitudes of the gene expressions. If magnitudes should not be considered, but only relative changes, the data should be mean centered to remove the dependence on magnitude. Mean centered data have a mean expression of zero, which is accomplished by subtracting the column (gene) mean from each data entity.
In most qPCR studies, interest lies in relative measurement where only differences in Cq values are important. Such data are usually better visualized and analyzed when mean centered. Mean centering also usually the preferred option when you are classifying samples based on variables that are all measured in the same unit.
For further examples and information about mean centering, see the tutorials Scaling during pre-processing and Scaling during processing on www.multid.se