If you intend to you a server for CNT structure, CHARMM-GUI has a nanomaterial modeller. The problem is it could only generate pristine single walled CNTs.
As regards, ssDNA docking on CNTs, I have found that AutoDock Vina can be misleading. I have used several proteins (cytoplasmic enzymes and membrane proteins) for docking with CNTs but the results have been misleading. My suggestion is to look for any other alternatives. I haven’t worked with DNA per se but expect such misleading results to occur in DNA-CNT systems as well.