Combined Twinning and Disorder

3/5) Fix disorder using FRAG...FEND in SHELXL

In the last section we incorporated twinning into the model and ran a few rounds of refinement. The result, copied from the end of the previous section, is shown below:

In the above image, the three largest difference map peaks provide the key to how this molecule is disordered. Note how peaks Q1 and Q2 are positioned just to the left of atoms O2 and C4, while peak Q3 is just to the right of C3. If we were to flip the whole molecule by 180° about a vertical axis in the plane of the image, the new positions for O2, C4, and C3 would coincide with difference map peaks Q1, Q2, and Q3. C2 and Q6 also appear to be related by the same operation. From the looks of it, such a 180° flip for the water molecules would superimpose O1w on O2w and vice versa, which means we don't need to model disorder for the water molecules. We therefore proceed by setting up for the next round of refinement with a fragment (FRAG) consisting of a whole uric acid molecule, and instructions to fit O2, C4, and C3 of the minor fragment to the positions of peaks Q1, Q2, and Q3. To make it easier to follow, we'll rearrange the atoms a little to place O2, C4, C3 at the top of the list. The edited file looks like this:

Ok, what to make of that! We'll break it down, step by step:
1) The FRAG line has code 17 (see SHELXL manual) and is followed by the unit cell dimensions. Here, the unit cell is needed because the fragment is given in fractional coordinates. If the fragment were defined by Cartesian coordinates, the unit cell would not be needed on the FRAG line.
2) Coordinates of the fragment atoms, in the same order as the main component. This is merely a copy of the uric acid atom lines, but with occupancies and displacement parameters removed (they're removed only for the sake of clarity, as SHELXL would ignore them if we'd left them in - see the SHELXL manual!).
3) FEND marks the end of the fragment.
4) A second number has been added to the FVAR instruction. It is used to define the occupancy factor for the major component, and it be will refined. The minor component is set so that the sum of occupancies equals 1.0.
5) As a precaution, strong SIMU restraints are applied to displacement parameters of the disorder components. If all goes well, we'll relax this later on.
6) The next 14 lines are the atoms of uric acid and water molecules. Note that the occupancies for uric acid are set to the second FVAR parameter, while the waters remain fixed at full occupancy.
7) The AFIX 176 ... AFIX 0 section defines which atoms are to be used in the rigid-body fragment fitting and which are just along for the ride. In this case, the first three atoms (O2', C4', C3') have been assigned the coordinates of Q1, Q2, Q3 difference map peaks. NOTE: You need to assign at least three atoms for this to work - bonus points if you can explain why! All other uric acid atoms are given zero coordinates. SHELXL interprets this as "keep the geometry of the fragment the same, but superimpose it on the major component by fitting the (three) specified atoms". That's the plan at least! Let's run a round of refinement and see what happens. The view in ShelXle should look like this:

The superposition has worked as planned, but the structure still looks a bit messy. Unrealistic 'bonds' have been drawn all over the place and the size of the ellipsoids for atoms that overlap are too dissimilar. The first of these problems will be fixed by assigning the major and minor disorder components to separate PARTs, as in the sorbose and whole-molecule disorder tutorials. The rest will be fixed by imposing a more appropriate set of restraints on the model prior to going anisotropic, adding hydrogens, and completing the refinement.