Normalization is used to correct for technical differences in sample intensities. Min/max is the recommended default methods; quantile normalization should be used only when it can be assumed that the overall distribution of phylotypes (the phylotype identities can vary, though) within each sample is approximately similar. These standard options correspond to different background assumptions:

• Min/max The default option. Aims to normalize the samples to comparable scales without affecting the phylotype distribution within a sample. Particularly useful when there are considerable systematic differences between samples, for instance due to antiobiotic treatment. Sets the minimum and maximum values of each sample to the same value by shifting and scaling the samples accordingly. These values are selected as the 0.5% 99.5% quantiles of the data to improve robustness.

• Quantile The quantile normalization forces the same phylotype distribution on all samples, which is calculated based on the average over all samples. Quantile normalization may be more efficient in removing technical biases from the data than min/max but it assumes that the overall phylotype distribution is approximately same for the different samples (ie. there are no antiobiotic treatments, and the samples/subjects are also otherwise similar).

• None skip normalization