Hi. MaxMedian is a variant of "median normalization", i.e. all medians per sample are equalized. But the medians in MaxMedian are set to the highest median of the samples. The rational is to have correction factors that are between [0,1] (or scaling factor above one [1,)) that can be used as QC. In similar vain, MaxSum is just "total intensity" normalization (sum of columns) and scaled to the maximum of the sums with similar reasoning.