Hi,

I have used Voom to transform my data with the TMM scaling factor and identified a list of differentially expressed genes for my RNA-seq data. What I want to do next, is to generate a bar plot on some genes of interest and also plot an expression heatmap. My question is, can I use the transformed E values from Voom for this, or should I recalculate the Log2CPM values using the cpm(x, log=T, prior.count = 0.5). I had taken a look at the transformed counts from both functions and their differences are subtle. I'm not sure which one is most ideal. Thanks for the help in advance.

For plotting and descriptive purposes we recommend cpm(x, log=TRUE). Note that you do not need to set prior.count=0.5, just use the default. See for example Section 5.1 of the limma-voom workflow.

The voom E values are designed to be used in conjunction with the voom precision weights. For plotting or downstream analyses that can't use the precision weights, it is better to use cpms with a larger prior.count in order to reduce the variance of the log2CPM values for small counts.