Figure 1
Data were collected on a slit-geometry instrument. Subsequently, all presentations are for smeared data and fits. Scattering data are typically presented as linear X-log Y plots (a) alongside the corresponding P(r) curve (b). A log X-log Y plot (c) is also acceptable as it emphasizes the low-angle data that carry the strongest signal and provide the most information regarding the overall shape of the molecule. A sample free from aggregate or interparticle interference will also be flat at small angles, again providing the reader with a rapid diagnostic of data quality. The linear X-linear Y plot (d), however, will obscure both the low-angle information as well as any fits made and must be avoided. Additional representations of the data include the Guinier plot (e) and the Kratky plot (f). The former provides a rapid diagnostic of sample quality, as deviations from linearity would be indications of either nonspecific aggregation (upturn) or interparticle interference (downturn). The latter provides information as to the folded state of the macromolecule: a fully folded protein would have a parabolic peak followed by convergence at a constant value at high q, while a fully disordered protein would show an increase at high q. If the Porod invariant is used to calculate the molecular mass of the solute, it is necessary to show the Kratky plot to demonstrate that the sample is folded and therefore that the calculation is valid. In this real example, the presented data were used for structural modelling of lysozyme and three orthogonal views of the models generated are presented (g). 12 DAMMIF calculations (Franke & Svergun, 2009BB3) were performed [a typical fit is presented in magenta in (a), (c) and (d); [chi]2 = 1.27] and averaged with DAMAVER (Volkov & Svergun, 2003BB24) to produce the averaged and filtered shape shown in magenta in (g). It is important to cite the mean normalized spatial discrepancy value and its standard deviation (in this case 0.507 ± 0.009) and whether or not any models in the set were rejected (in this case none) to quantify the degree of similarity among the models generated. In this example, the total volume occupied by the spread of all of the models (aligned for maximum overlap) is shown in grey, with the most-populated volume presented in magenta. The crystal structure of lysozyme has been superposed (black cartoon) on the dummy-atom structure with SUPCOMB (Kozin & Svergun, 2001BB13) and its fit to the scattering data calculated with CRYSOL [black line in (a), (c) and (d); [chi]2 = 1.56; Svergun et al., 1995BB23]. Sources for the discrepancy in the fit for the high-q data should be considered in comments on the interpretation of the data. With the data presented as above, it is possible to see that there is a small upturn at high q in the Kratky plot (e), which may be indicative of flexibility (unlikely in the case of lysozyme), a difference between the internal structures of the model (e.g. high-resolution features not fully accounted for) and the measured data, or a poor solvent subtraction. The P(r) curve would support the poor subtraction possibility, as the curve does not cleanly approach zero at r = 0. With these data available, a reviewer may recommend that the experimenter repeat the measurement before publication, depending on the interpretations made in the manuscript.  [article HTML]

© International Union of Crystallography 2012