Adding all phosphates to flexibility parameter will increase the computational time. Moreover, what will be the point to include flexible residues that are far away from your ligand that will never interact with your compound? Instead, I think you should prefer bases (or nucleotide residues) that are in near vicinity of your ligand should be considered as flexible. To know in advance what bases are supposed to interact with your compound, I suggest you to use DNA model from PDB that is already in complex with some ligand. This will serve three purposes 1. you will know the exact binding site for ligand. 2. you can easily define flexible residues (residues that are, say, in 3.0 angstrom radius of crystallized ligand) and 3. you can use this complex to validate the docking system and fine tune parameters required for docking run by re-docking the experimental ligand back into DNA and see if it predicts the crystallographic conformer.
Adding all phosphates to flexibility parameter will increase the computational time. Moreover, what will be the point to include flexible residues that are far away from your ligand that will never interact with your compound? Instead, I think you should prefer bases (or nucleotide residues) that are in near vicinity of your ligand should be considered as flexible. To know in advance what bases are supposed to interact with your compound, I suggest you to use DNA model from PDB that is already in complex with some ligand. This will serve three purposes 1. you will know the exact binding site for ligand. 2. you can easily define flexible residues (residues that are, say, in 3.0 angstrom radius of crystallized ligand) and 3. you can use this complex to validate the docking system and fine tune parameters required for docking run by re-docking the experimental ligand back into DNA and see if it predicts the crystallographic conformer.