Hi @orangeboomerang, this can sometimes happen if the “symmetry alignment” that happens at the start of the refinement with “O” symmetry fails to correctly align the starting structure.
If you have any other apoF structure that is already symmetry aligned, you can try using that as an initial model. Or, you can download the C1 map and in Chimera, rotate and re-output it to be aligned to the symmetry axes, then use that as the initial model.