How to choose a suitable normalization procedure for my gene expression data?

If the data is not already normalized, you need to choose a proper method for data normalization. It is generally considered that differences in expression exist on a multiplicative scale. For microarray data analysis using linear model (limma), log transformation should be used to bring them into the additive scale, where a linear model may apply. For RNAseq data analysis using edgeR, the Trimmed Mean of M values (TMM) should be used. Many common normalization methods have been offered for QPCR data normalization. Point the mouse to the help icon on the normalization section to find more about each normalization procedure.

If you are not sure whether the data is already log transformed or not, you can easily figure this out by visualizing the data (i.e. boxplot). For microarray data, log transformed data values are usually less than 16. For count data with 1 million counts, log2(1,000,000) is less than 20. Therefore, if all data values are below 20, it is reasonable to assume that the data has already been log transformed.