Hello,

I am working with TCGA gene expression data (microarray based) level 3 for breast cancer and I have seen that the majority of the expression values is negative, even in genes that are known to be overexpressed. From what I understand the expression values are calculated as the log ratio Cy5(sample)/Cy3(universal human reference), so all the genes that are overexpressed should be above 0.

Could anyone tell me what these negative values mean? Am I missing a normalization step?

Thanks!