I am new to the field and wonder what the differences are between edgeR and limma as both are recommended for RNA-seq and other applications. Tried reading the papers and manuals but my stats background is not sufficient to really understand all concepts in detail.
- Why for example is it possible to analyse the same data with two different models, either a linear or a negative binomial one?
- Why exactly is limma so much faster, does this has to do with the models`?
- Does limma estimate a dispersion as edgeR does? Because this seems to be the part that in edgeR takes very long when you have many groups or large n. Why is this faster in limma?
- Which part in the edeR workflow matches the eBayes step in limma?
- given that limma works for RNA-seq and scores well in the benchmarks I read why did the authors actually develop edgeR (no offense, really just wondering)?
Thank you for your time!