Combined Twinning and Disorder

4/5) Add restraints/constraints to enable stable refinement

In the previous section we added a minor disorder component to the model using FRAG...FEND and ran a few cycles of refinement. The result, copied from the end of section 3, is shown below:

All that remains to be done is to assign the major and minor components to separate PARTs, add hydrogen atoms, and refine anisotropic displacement parameters for the major component. For the minor disorder component, there is little justification for going beyond the isotropic level because the occupancy fraction is so small. (It can be done, but requires special treatment - see section 5 if you are curious.)

To proceed, we'll edit the *.res file to insert the required PARTs, make the major component anisotropic, and throw in a few restraints to ensure that the model continues to make chemical sense throughout subsequent refinement runs.

In the above file, the SIMU restraints have been removed for the major component and relaxed considerably for the minor. PART numbers have been assigned to both uric acid fragments, a SAME restraint has been added, which should ensure that the minor component geometry closely matches that of of the much better defined major component. Finally, the major component is set to gain anisotropic displacement parameters. The RIGU restraint here is just a precaution: it may or may not be needed, but it does no harm. As a later exercise, you could try removing this RIGU to see what happens. The minor component displacement parameters are still constrained to be isotropic. When refined with SHELXL, the result looks like this:

The refinement is looking good. The largest difference map peaks look to be consistent with hydrogen atom positions on the uric acid nitrogen atoms and on the water molecules. Since their geometry is well known, ShelXle should be able to add the N-H hydrogens automatically, or you could add them using HFIX 43. Since major and minor PARTs have slightly different hydrogen atom positions, it would be unwise to try to refine these hydrogens freely. There is no convenient way to automatically add the water hydrogens, so they will have to be assigned using difference map peaks. To be on the safe side, we'll add the uric acid hydrogens first, refine, then add the water hydrogens, perhaps with O-H distance and H-O-H 'angle-distance' restraints. Not shown here, the water molecules should be set to PART 0, as they are not disordered. Once all the atoms are included, the weighting scheme can be optimized and other minor things such as testing for extinction can be performed. The end result after all these tasks should look something like this:

The refinement should now satisfy even the most pedantic referee. The final R-value is 3.78%, the difference map is essentially flat (largest features are +0.167 and -0.137 eÅ^-3), the weighting scheme looks normal, etc. All that remains is to create and edit the CIF. Having said that, it is possible to make a few tweaks to the model that will reduce the number of parameters needed, albeit by adding a substantial degree of complexity. See below for section 5 if you are interested.