I am trying to generate a gene signature using data from affy 3' chip. Most of the reported gene expression based classifiers use log-intensities to generate a gene signature. Is there any reason for that? Can we also use log-ratio (such as cancer1-normal,cancer2-normal, etc...) instead of intensity. The gene expression intensity should be analyzed in relative-terms rather than absolute value, is that correct?
Your reply will be really helpful.