I have quite a strong batch effect in my data (paired design,i.e. first replicates of treated and control were done together and second replicates of treated and control also together). Additionally, I need to say that I am studying quite a moderate effect so ,that I expect a small number of genes to be DE.
To correct for it I used RUVseq, however, I am not sure if I better use RUVs or RUVg from the RLE and PCA plots I got.
Since you already know beforehand what the batch effect is, I think you'd be better off to just add a factor for it in your linear model while running your differential expression analysis instead of going the RUV route.
According to my understanding, there is no need to correct for batch effects in a paired analysis. But I do not know if RUVSeq would be beneficial or introduce biases in the paired analysis.
