difference between voom and calcNormFactor normalization methods for RNA-Seq analysis

Question

[upendrakumar.devisetty]

What exactly is the difference between between voom (limma) and calcNormFactor (edgeR) normalization methods? Though it is not very clear to me, I understand the "voom" normalization is better, but in some articles I see that people use both normalizations simultaneously. i.e calcNormFactor followed by voom and then do limma based linear modeling for differential gene expression analysis. If voom is better than calcNormFactor, why are people using both of them together? Can anyone shed some light on this please.

[fabio.ticconi]

Have you found an answer somewhere, by any chance?

[ddiez]

This answer to a similar question in the bioconductor support site might be relevant: https://support.bioconductor.org/p/77664/#77666

And also, make sure to read the next answer (and comments) in that thread about using both TMM and quantile methods: https://support.bioconductor.org/p/77664/#77665

Answer 1

[Gordon Smyth]

voom is a analysis method, not a normalization method. Your question is a bit like asking: what is the difference between an engine and a steering wheel? The answer is that they are designed to be used together. You can choose between different engines and you could choose between different steering wheels, but you usually want to have one of each.

[ddiez]

Hi Gordon. I understood the OP was asking about the difference between using normalize.method argument in voom() v.s. using calcNormFactor()- hence my suggestion above. Now it is not so clear what the question was- maybe my understanding was biased due to me wanting to know that difference when I found this thread...