As you may know "bonds" are not really physical. More to the point, ATK doesn't use bonds in the calculations. sp3 or sp2 hybridization is a useful picture when employing simpler models like tight-binding, but doesn't play any real role in DFT. In the Huckel model we do have different parameters for sp2 and sp3 hybridized carbon, for instance, but the bonds between atoms in the structure are still not used in the computational model.
Still, it's often useful to see bonds to get a better picture of the structure, and this will be included in the next version of ATK (11.2). But it will just be based on a geometric construction, using the covalent radius of each atom as a bonding criterion.
One way to see if the molecule hybridizes with the electrode would be to look at the electron density in the interface region. One could also compute the Mulliken overlap populations, but that's a bit harder.
4 microamp at 2 V sounds normal, why do you think it's wrong?
