Optimize 3d pdb cannot run haddock

Previously, I run haddock using ligand extract from bound crystal. However, the docked result is not the same as reference crystal.

So now I try to optimize geometric my 2D structure that I download from pubchem and convert to 3d using avagadro. However the pdb produce missing some column, and cannot submit to Haddock for docking.

There I would like to know is it correct for me to just extract the ligand from bound crystal? Do I need to do any refinement? And is there a way to convert 2D to 3D without the missing column in PDB? Would appreciate if anyone can kindly advice.

In terms of conversion from smile to 3D, we like the omega toolkit from OpenEye (free for academics). But there are indeed other solutions.