Docking with DNA is a bit tricky. I won't make the DNA flexible. I would go for MD instead. In DNA pdb files base+sugar+a phosphate is considered as a unit (nucleotide). That is why you could not select the base and sugar separately.
Besides, in order to predict intercalation, you need to take a ligand intercalated DNA structure. Because, rise or other base pair geometry changes cannot be modelled by simple docking.
I do not have fast method. For me, I have to figure out the name of sugar residues with PDB file. And in PMV select from string. That is my way to choose them separately.
Since I am much familiar with VMD, I usually use VMD to find the name of specific residues. And use the residue name in PMV.