Switch to additive scale

Expression levels can be signal intensities from oligonucleotide chips or 'red-green' ratios from cDNA chips. You will be used to compare expression levels by fold changes. This amounts to operating on a multiplicative scale. For statistical analysis, the multiplicative scale is not convenient and you should transform the data to the additive scale by taking for example logarithms. On the additive scale, differences between values correspond to ratios on the original scale. For example, the difference between two log expression values relates to the fold change on the original scale. Several normalization methods already return the data on the additive scale, for example the variance stabilization method by Huber et al. (3). If your data is not on an additive scale, type X <- log (X). If your data has negative values, refer to a text on expression data normalization, for example Huber et al. (4).

0 0

Post a comment