was referring to the partial charge in the restraint file, which IMHO should be translated to the model.

I have a second problem - I'm using the cif file for refinement but the angles and bond lengths still get distorted. These are flagged as outliers in phenix.refine. Why is it not using the values that I input? My resolution is 3.8 A, but I still end up with positive density around all the hemes. In fact, the problems with hemes are the cause of a large majority of my outliers. I did a parallel refinement with and without the restraint file - it doesn't make any difference.