Automatic protein structure solution from weak X-ray data

Abstract

Determining new protein structures from X-ray diffraction data at low resolution or with a weak anomalous signal is a difficult and often an impossible task. Here we propose a multivariate algorithm that simultaneously combines the structure determination steps. In tests on over 140 real data sets from the protein data bank, we show that this combined approach can automatically build models where current algorithms fail, including an anisotropically diffracting 3.88 Å RNA polymerase II data set. The method seamlessly automates the process, is ideal for non-specialists and provides a mathematical framework for successfully combining various sources of information in image processing.

Figure 1. The current and new combined approach for structure solution.

(a) Currently, when solving a structure using anomalous scattering, the steps of experimental phasing, density modification with phase combination and model building with refinement are performed separately. (b) Unlike the traditional stepwise approach, the combined function simultaneously uses the information from density modification, model building and from the data to provide the best estimate of the electron density.

Figure 2. Detailed diagram for the combined algorithm.

An expanded UML flowchart of the combined algorithm, which includes decision making (blue diamonds) to skip the model building.

Figure 3. Comparison of CRANK's stepwise and combined approach.

The fraction of model correctly built by the CRANK's stepwise approach compared with the new multivariate combined method on 147 data sets. Each data set is represented by a circle. The y axis plots the fraction of model correctly built using the combined algorithm, whereas the x axis shows the performance of the stepwise traditional algorithm. The further a circle lies above the dotted diagonal line, the greater the improvement the new approach provides.

Figure 4. Comparison of PHENIX and CRANK’s combined approach.

The fraction of model correctly built by PHENIX compared with the new multivariate combined method on 147 data sets. Each data set is represented by a circle. The y axis plots the fraction of model correctly built using the combined algorithm, whereas the x axis shows the performance of PHENIX. The further a circle lies above the dotted diagonal line, the greater the improvement the new approach provides.

Figure 5. RNA polymerase II electron density.

Electron density of a portion of the 3.88 Å RNA polymerase II structure automatically built by the new combined approach contoured at 2.1σ. The final Cα trace is shown in grey, whereas the automatically built model is multicoloured. This figure was made with COOT. The length of the scale bar is 5 Å with minor ticks at 1 Å.

Figure 6. Deposited and automatically built clamp domain–Spt4/5 models.

(a) The final, deposited structure of the related RNA polymerase clamp domain in complex with Spt4/5. (b) The automatically built structure using the combined method of the related RNA polymerase clamp domain in complex with Spt4/5. Fig. 6a,b were made with MOLSCRIPT.