should I use normalized counts or transformed of normalized counts for RNA-seq association analysis?

Question

[Alewa]

Hi,

I run differential gene expression with DeSeq2.

dds <- DESeq(dds)
rld <- rlog(dds, blind=False)
vsdata <- vst(dds, blind=FALSE)

If my goal is to find association between gene expression (RNA-seq ) and methylation, should I use the normalized DESeq object dds or transformed normlaized counts rld / vsdata or the log2FC? previous study use FPKM, RPKM

I'm confused & thanks in advance for your help

Answer 1

[Trivas]

Depending on your study design, I think the best you'll be able to do is say whether the differentially expressed genes from DESeq also have differential methylation patterns from whatever software you're using, grouping it as +/+, +/-, -/+, or -/- as an example. You could also gate either DEGs or methylation status and do a metagene analysis (e.g., histogram of read depth over certain regions). See here for a very simple analysis https://www.frontiersin.org/articles/10.3389/fpls.2021.741415/full#F5

[Alewa]

Thanks for the reference!!!!!