I think you may still try doing the Molecular Projected Self-Consistent Hamiltonian (MPSH) analysis (described in the tutorial) for the molecular junction at nonzero bias voltages. Calculating Projected Device Density of States would complement this analysis, as you might also be able to see some PDDOS peaks related to molecular levels.