Automated MAD and MIR structure solution

Abstract

Obtaining an electron-density map from X-ray diffraction data can be difficult and time-consuming even after the data have been collected, largely because MIR and MAD structure determinations currently require many subjective evaluations of the qualities of trial heavy-atom partial structures before a correct heavy-atom solution is obtained. A set of criteria for evaluating the quality of heavy-atom partial solutions in macromolecular crystallography have been developed. These have allowed the conversion of the crystal structure-solution process into an optimization problem and have allowed its automation. The SOLVE software has been used to solve MAD data sets with as many as 52 selenium sites in the asymmetric unit. The automated structure-solution process developed is a major step towards the fully automated structure-determination, model-building and refinement procedure which is needed for genomic scale structure determinations.

Figure 1

Steps in automated structure determination by SOLVE.

Figure 2

Determining the heavy-atom partial structure in SOLVE.

Figure 3

Z scores for one model structure determination. Each point correponds to one trial heavy-atom partial structure. The x axis is the quality of the solution (the correlation coefficient of the map calculated using the trial heavy-atom structure with the true map). The y axis is the Z score for the scoring criterion. The scoring criteria shown are (a) agreement with the Patterson function, (b) the cross-validation difference Fourier analysis, (c) the figure of merit of phasing and (d) the distinction between solvent and protein regions in the native electron-density map.

Figure 4

Probability of identifying the better of two trial heavy-atom solutions. Automated structure solutions were carried out on 419 model data sets. For each trial heavy-atom solution, the individual Z scores for each scoring criteria and the overall overall Z score were noted. The quality of the map, based on the correlation coefficient of the native electron-density map to the model map, was recorded as well. All pairs of heavy-atom solutions for a single model data set which differed in quality (correlation coefficient) by 0.05 ± 0.02 were then examined to determine whether the solution with the higher Z score had the higher correlation coefficient. The percentage of cases in which the Z score correctly identified the solution with the higher correlation coefficient is plotted as a function of the correlation coefficient of map to model map obtained from the solutions.

Figure 5

Highest Z scores and quality of electron-density maps for 419 model structure determinations. Each point correponds to the highest scoring solution from one model structure determination. The x axis is the correlation coefficient of the map calculated by SOLVE with the model (true) map. The y axis is the overall Z score for this solution.

Figure 6

Identification of hand of selenium partial structure using the native Fourier. (a) Section through an electron-density map of β-catenin calculated using 11 correct selenium sites. (b) As (a), but with inverted hand of Se atoms.

Figure 7

SOLVE electron-density map of β-catenin.

Figure 8

SOLVE electron-density map of Rhodococcus dehalogenase.