When batch correcting gene expression data one could either do a two step procedure (e.g. Combat + DE analysis) or in a single step where batch is incorporated into the stats as a covariate.
This discussion (Batch effects : ComBat or removebatcheffects (limma package) ?) and the paper it references indicate the second way is preferable.
However the first method had the benefit of allowing a 'corrected' dataset to be assessed using PCA, clustering.
Is there any way (e.g. in DESeq2, edgeR or Limma) to do a similar assessment of batch correction sucess using the single-step method?
Thanks in advance