Model building crystallographic refinement

Structure Refinement. The refinement of the structure was based on an energy function approach (Brunger etal., 1987) arbitrary combinations of empirical and effective energy terms describing crystallographic data as implemented in XPLOR. Molecular model building was done on an IRIS Workstation (Silicon Graphics) with the software TOM, a version of FRODO (Jones, 1978). 

A set of atomic coordinates is obtained from X-ray crystallographic or NMR structure data, or by model building. The structure is first refined to relieve local stresses due to overlaps ofnon-bonded atoms, bond-length and angles distortions, etc. 

At this stage, assuming that the nominal resolution is 4 A or better, the electron density map should be of sufficient quality to interpret readily the course of the polypeptide chain and rapidly build a model, usually for the icosahederal asymmetric unit. Conventional crystallographic refinement techniques are then employed to refine the model against the observed data. 

Johnson responded with his opinion that in the past most protein crystaDog-raphers believed they could not do refinements profitably, but that techniques are changing and that the field will soon reach a state where the precision of the geometrical parameters can be evaluated objectively. WILLIAMS asked just how much does the protein crystallographer rely on results of small-molecule crystallographic structures, and JOHN SON replied that heavy reliance is placed on these results in model building. 

The basic sequence of steps in the crystallographic analysis of a hyperthermophilic protein is unaltered from that used for more conventional proteins one must (1) obtain sufficient quantities of pure material (2) produce diffraction quality crystals (3) determine the phases and finally (4) complete the structure determination by building and refining a molecular model. Increasingly, the rate-determining step in any structure analysis, including hyperthermophilic proteins, involves obtaining suitable samples for crystallization trials. 

Crystallographers have had some success in using normal mode descriptions of internal motion, with effective frequencies as adjustable parameters, and initial steps have been taken to incorporate similar ideas into NMR refinements. This model builds in important large-scale correlated motions, but not generally the effects of local conformational Jumps. 

---