I have looked at the reference, and I do not think it is surprising that the structures are a little different. Whether that difference in the structure is important for what you want to model is up to you to decide.
As mentioned above, we introduce some strain into the structure when it is forced to a cubic lattice, and it is probably the relief of that strain you see. If you wish to reproduce exactly the calculations in the paper, you need to construct the nanowire directly from the rhombohedral crystal structure, which is unfortunately currently not possible in VNL.