Part 4: Further Exercises - try one or more of these
A. Self analysis
Choose one of your own structures, or one you're especially interested in, and analyze it in MolProbity.
One possibility would be to look at initial chain tracings of the gold lysozyme. The warpNtrace pdb file is in /home/macroxta/tutorial_data/coot/au_lys/, and you should have a saved model from your own work in Coot. [In general, the parts successfully traced by these modern programs are very good quality (esp. at high resolution) with excellent rotamers & Ramachandran and relatively few clashes.] Upload te pdb into MolProbity and add H: usually there will be some NQH flips, since the tracing only used the density; look at them in the flipkin, and remember that you can have these corrected easily each time you build or extend a model. Evaluate geometry and contacts, and either look at the multi-criterion kinemage or save the Coot to-do-list and look at problem clusters with contacts calculated in Coot. The warpNtrace model has an Arg stuffed into the gold, a backwards Trp, a Gln with messy density that was put in an impossible place . . .
Read the .kin.gz file into KiNG, choose the view for L Arg 6, and read in the map using Tools > Structural biology > Electron density maps (note that it's in CCP4 format). Both Arg and G base fit their density well, and the guanidinium stacks nicely with the U above it (green vdW dots), but it seems to be trying, and failing, to make Arg-G H-bonds. Choose Tools > Structural biology > Sidechain rotator, select the pdb file (be sure to use the 1s72H one with H atoms, not the plain 1s72 one) and pick an Arg atom to enable fitting.
Right-click on the Cb and look down the Ca-Cb bond to judge the density placement, which clearly requires a minus chi1 (putting Cg opposite the backbone CO). Go back to the standard L Arg 6 view, scroll down in the rotamer list, and quickly try each one that starts with "m", and look for one that places the guanidinium somewhere close to the right position and orientation but flipped over about 180° from the original (don't worry much about clashes at this stage). Starting from the one rotamer that's promising, adjust the chi dials to get 2 H-bonds and a good density fit. There is still a small clash when the H-bonds are good; that level is not very worrisome, but you can improve the interaction using a small backrub change if you like. Now you have a classic double-H-bond Arg-G specific interaction. Accept your refit.
There are several other misfit Arg/RNA interactions in here (they understandably worked harder on the RNA than on the proteins), but only one other within the map provided. If you like, go to the C Arg 246 view, take the contour level down to about 0.9σ, and try refitting this one. Or, just explore the all-atom contacts for these protein/RNA interfaces.
C. Why Backrubs are therapeutic - a buried Ile in 1mo0
Procedurally, this example is the same as that of Thr 77 in Part 2. For Ile A120 in 1mo0, selection of the correct rotamer is difficult: the map does not indicate a clear choice and neither do probe dots. The key idea behind the Ile A120 rotamer selection is that one of the rotamers (pt) gives clashes on only one face of the sidechain. The backrub tool then allows one to finish the correction from the pt rotamer.
D. Cross training
Try KiNG on 1sbp Thr 32 or other problem residues.
Try the various validation tools in Coot on your saved 1bkr file, and see which ones show up which problems. You fixed Thr 77, but Thr 101, Arg 32, and the chain termini should show up (each in at least one tool, but not in all). Note that to fix Arg 32, you need to delete water 1075 (under Calculate: Model/Fit/Refine). At the N-terminus, you can find density to add another peptide (again, deleting a water), by "Add terminal residue" and picking the old N atom. Asn 23 is a bad rotamer in MolProbity if not already flipped by Reduce; it's an easy and satisfying fix in Coot.
Try Coot on rebuilding the arginines in 1s72 or in 1bkr. See if you can get Coot to flip 1bkr His 42.
[Note that Coot needs CCP4-format maps (or a .mtz file); we've provided a ccp4 map for the 1s72 ribosome piece, used also in KiNG.]