I have a very simple doubt and i know the topic has been discussed over and over but still not confident enough to choose the right approach. I am analysing my proteomics data by using RAW intensity values. Do i have to do imputation followed by normalisation or other way around. Because when i normalize (VSN) the data it looks fine but as soon as i impute (Miniprob or KNN or global min) the data, my meanSDplot kind of dispersed. So i really don't know how to convince my supervisor. What will be the right approach to do? What i am thinking is that by imputing after the normalisation, i am introducing the variance which may give false positive am i right?