At what step should I perform batch correction?

Question

[prithvi.mastermind]

I have extracted RNA-Seq and microarray datsets and interested in performing batch correction to my data. I have some queries in my mind which I'm struggling to solve:

1) Which batch-correction method is best so far? Noiseq or MultiBaC or Comat-Seq?

2) When should I actually perform batch correction on my data? After normalization or before normalization or during DEGs identification?

[ATpoint]

Please post details, what is the experimental design and what are the groups to test. Can well be that you cannot correct for anything depending on what you want to test (confounding). Combining completely independent technologies is usually hard or impossible (or at least not meaningful).