I used the tximport pipeline upstream quantification methods (Salmon) in order to further use EdgeR pipeline. I am confused about normalization part. As in the (https://bioconductor.org/packages/release/bioc/vignettes/tximport/inst/doc/tximport.html#edgeR) the Computing effective library sizes from scaled counts was done by this code <eff.lib &lt;-="" calcnormfactors(normcts)="" *="" colsums(normcts)&gt;="" and="" it="" is="" written="" that="" y="" &lt;-="" y[keep,="" ]<="" p="">

y is now ready for estimate dispersion functions see edgeR User's Guide

So my question is should I normalize again using TMM method or not?????????? Could you please explian this point.

Thanks in advance

The idea for any vignette is to show how to do something. It would be pretty odd for Mike Love to show how to do something and completely leave a step out, no?

But to be clear, it doesn't matter if you 'normalize' again using TMM or not. The offsets you have inserted into the DGEList object take precedence over the library sizes, which you can see in getOffset from edgeR:

> getOffset
function (y) 
{
    offset <- y$offset
    lib.size <- y$samples$lib.size
    norm.factors <- y$samples$norm.factors
    if (!is.null(offset)) {
        return(offset)
    }
    else {
        if (!is.null(norm.factors)) 
            lib.size <- lib.size * norm.factors
        return(log(lib.size))
    }
}
<bytecode: 0x000000001c0d5ef0>
<environment: namespace:edgeR>