Accurate diffraction data integration by the EVAL15 profile prediction method : Application in chemical and biological crystallography

Abstract

Accurate integration of reflection intensities plays an essential role in structure determination of the crystallized compound. A new diffraction data integration method, EVAL15, is presented in this thesis. This method uses the principle of general impacts to predict ab inito three-dimensional (x,y,omega) reflection profiles, with only a few physical crystal and instrument parameters. By least-square fit of the predicted profile to the observed profile using Singular Value Decomposition, the net intensity of a reflection is obtained. The detailed profile analysis has the additional advantage that specific physical properties of the crystal are revealed. The EVAL15 method is particularly useful in circumstances where other programs fail, such as regions of reciprocal space with weak scattering, crystals with anisotropic shape or anisotropic mosaicity, K(alpha1),K(alpha2) peak splitting, interference from close neighbours, twin lattices or satellite reflections of modulated structures, all of which may frustrate the customary profile learning- and fitting procedures. EVAL15 straightforwardly allows the deconvolution of overlapping reflections. This thesis shows that the profiles can be predicted satisfactory and that accurate intensities are obtained. It’s application on both standard and overlap diffraction data is investigated and the quality of EVAL15 data is assessed and compared to that of EVAL14 (Duisenberg et al., 2003), where predicted contours are used in the summation integration of reflections, in particular to investigate if the EVAL15 profile method results in improved quality of the weaker data. We found that EVAL15 delivers a lower Rmerge, a higher I/sigma (especially in the highest resolution shell) and a lower R-value in the refinement. The EVAL15 data are also more successful in phasing. The deconvolution of overlapping reflections result in higher completeness and redundancy of the data and contribute to improved electron density maps. Duisenberg, A. J. M., Kroon-Batenburg, L. M. J. & Schreurs, A. M. M. (2003). Journal of Applied Crystallography 36, 220-229.