Hi Michael and community,
Thanks for the work and efforts on DESeq2! I've had this question for some time, I went through quite a number of posts, the manual, and the vignette, but can't really find the answer. What is the unit of the "normalized count" of the y axis of the "plotCounts" plot?
plotCounts(dds, gene=which.min(res$padj), intgroup="condition")
in the vignette.
Question1: If I type ?plotCounts in R, then it says "Normalized counts plus a pseudocount of 0.5 are shown by default." and also there's this "normalized = TRUE", and "transform = TRUE" arguments. So is the y axis just raw read counts divided by DESeq2-estimated size factor? or has it been logged? (loge? log2? log10?) Also, why is the psudocount 0.5 here? (it's just to avoid taking log of zero, right? but why don't you set it smaller, say, 0.001?) What happens if we set normalized = FALSE? Likewise, what happens if we set transform = FALSE?
Question2: This one is indirectly related to the first question. So with the way DESeq2 normalizes raw read counts without the effect of gene size, can we compare the expression of two (or multiple) different genes within the same sample? If not, is there other recommended way of normalization for this purpose?
Thank you in advance!
ps. R version 3.4.3 (2017-11-30), DESeq2_1.18.1