TMVA: add weighted variable

Dear experts,

I want to build a BDT with several variables out of which only one needs to be reweighted. So if I have in the ntuples the variables (a,b,c,d) I want the tmva to train and test with (alpha, b, c, d) where alpha is the reweighted ‘a’.

I have seen you can add a weight to the full signal or bkg sample, but it looks like there is no way to reweight an individual variable, is that true?

thanks
marc