I realize that when correcting for GC bias via EDAseq and doing DE in DESeq2, the recommended method is to pass EDAseq's offsets to your DESeq object as NormalizationFactors as per the DESeq2 manual. However, we have a previous analysis that was done a few years ago, where the EDAseq normalized counts were passed directly into tximports' count slot prior to creation of the DESeq object. While I know this is not the current recommended pipeline, is this completely wrong or just less ideal (I know this was recommended in original DESeq prior to NormalizationFactors)? I only ask because we have since validated a number of these targets that are significant in the old analysis but are not when doing it the recommended way.
Thanks, I appreciate your time!