in these cases, both HA and IgG are background controls. Theoretically,
both of them should have no pulldown signals at all (= 0). If you divide
something by HA or IgG signal, any little variation will change your ratio
dramatically. So, neither of them is the right normalization.
A good normalization method is using 1% input as a control.
However, the best normalization is using some internal controls that are not
changed by your treatment. For example, your TF binds gene GAPDH and gene
X.