thanks for the prompt reply. The command-line Melodic runs just fine except that I don't get the right dimensions (at least not the ones I requested) in the output. The question I had about the registration was about how residualized data might behave in the GUI, especially if the grand mean, CSF and white matter have all been regressed out.
Also do you have any thoughts on the other issues?
Again I really appreciate your assistance with this.