In normalization method, geometric mean is obtained across the all samples regardless of control and experimental group, and then each data is divided by the geometric mean.
1. I really want to know what "data/geometric mean" means.
2. Why is geometric mean (not other values like arithmetic mean) used, depite samples are supposed to be independent (no relation between samples) and there is no extremely big differences in counts between samples in most biological data?
3. For DE test, is EM implemented based on negative binomial distribution or beta distribution? and why?
Thanks for you help in advance.