The graphs you show are - most likely - for a simple model calculation with a single channel (it looks like just a square barrier transmission).
The transmission values reported in ATK are already normalized, in the sense that they are presented in units of the conductance quantum. The fact that T(E) may be larger than 1 is not a problem of normalization, it's a reflection of the fact that there may be more than one channel present at a particular energy. Thus normalizing it to unity would be quite misleading, and in fact wrong.
