As shown in the attached image, the modified W is far off from a normal C. There is no N1 atom. Changing N2 to N1 does not help because N2 in W is connected to C4 and C6, not C2 and C6 as is the norm for N1 in pyrimidines. Thus, up to 3DNA v2.1 released on 2013jul10, W is not recognized a nucleotide, as can be verified by the following command:
find_pair -s W.pdb stdout
W.pdb
W.outs
1 # single helix
0 # number of bases
1 1 # explicit bp numbering/hetero atoms
So the reported behavior is as expected in the current 3DNA distributions. I will check if it is feasible to refine 3DNA's definition to include such modified cases in future releases. Alternatively, you may want to try the NUPARM/NUCGEN programs from the Bansal lab.
Xiang-Jun