This is a difficult question to answer. It varies based upon when you add the spike-in sample, how many different antibodies you are adding, how common you expect the binding to be, how different you expect the binding to be between the spike-in sample, and the samples you are analyzing, and more. In general, the aim is to have about 10% of your reads in your control condition be spike-in data. Too much spike-in and you won’t have enough data about your sample of interest, too little and the scaling becomes driven by noise rather than actual differences in global binding. – answered by Tovah Markowitz, Paul Schaughency, Vishal Koparde.
Please login to submit your answer