1 Department of Chemistry, Faculty of Natural Sciences, Constantine the Philosopher University, 94974 Nitra, Slovak Republic
2 Department of Inorganic Chemistry, Faculty of Natural Sciences, Comenius University, 84215 Bratislava, Slovak Republic
Density modification (DM) is a standard routine for improving phases, particularly for structures beyond atomic resolution . For atomic resolution structures the Fourier recycling with peak picking is usually sufficient tool to solve the structure. In cases the phases are not very good, Fourier inversion of modified superposition or electron density maps may be used as the first step of a phase refinement [2,3]. Density modification using molecular fragments (DMF) offers a tool for strengthening structure determination methods in a way analogous to molecular replacement. Chemical information can be entered into the process of structure determination by 6-D phased rotation and translation function (PZO method) .
Fragment electron density is calculated for top peaks of the PZO search. For each peak the fragment atoms are positioned in the unit cell. Electron density is constructed for FFT calculation of structure factors . Electron density is sampled on a uniform grid. Contributions from all fragment atoms are summed at each grid point. An electron density map belonging to one fragment and one PZO peak is created. Maps from different peaks and from different fragments are combined together to form a fragment map. In the process of density modification the fragment map is combined with electron density map. DMF map is modified by omitting low-density (negative) regions. Then structure factors are calculated by reverse FFT. Density modified structure factors (FDMF) are fitted against observed structure factors or normalized structure factors in bins of different resolution. These structure factors can be used for R-factor calculation and for weighting Fourier coefficients. The density modification (DM) is exactly the same procedure as DMF but no fragment map is used. Only low-density regions are omitted.
In the first group of tests organic structures containing sulfur or chlorine atoms were selected. Atomic minimum superposition was not used in the phasing process. Phases calculated from heavy atoms were used for density modification. Phases from heavy atoms were pre-refined by several cycles of DM, where necessary. Results of DMF were compared with results of DM. In every case the DMF is superior do DM. In both methods the cut-off was fixed at 4% of the top electron density.
Comparison with similar calculation on the same structure revealed, that the fragment positioning is an important step in this structure determination or a way of considerable reducing number of Fourier cycles needed to solve the structure.
 Cowtan, K. D. & Mail, P. (1993). Acta Cryst. D49, 148-157.
 Simonov, V. I. (1982). Computational Crystallography, edited by D. Sayre, pp. 150-158. Oxford, Clarendon Press.
 Shiono, M. & Woolfson, M. M. (1992). Acta Cryst. A48, 451-456.
 Pavelcik, F., Zelinka, J. & Otwinowski, Z. (2002). Acta Cryst. D58, 275-283.
 Agarwal, R. C. (1978). Acta Cryst. A34, 791-809.