I have a dataset with lots of dN/dS values coming out from PAML, my question is how can i normalize the data for parametric correlations? Most of values fall between 0 - 0.1.
There was a very similar question yesterday: Normalising tag count to RPKM. The input data were different but both boil down to the same misconception: that normalization is always required. For calculating correlation it is irrelevant if the data is normalized (as long as the normalization function is monotonous). So the answer is no normalization is required, and non-parametric correlation might be preferred, because the data might not be normal.
