MAD phasing guidelines

Preparation of diffraction data

• Use all data.
• If possible, use the Wilson statistics to put the MAD data sets on an absolute scale. This is important because phase refinement sometimes fails if the scale factor is very different from one. If Wilson statistics cannot be used because the high resolution limit of the MAD data is not at least 3 Å, a high resolution native data set should be used if available. Use the high resolution data set for the determination of the absolute scale factor, and scale the MAD data sets relative to this one.

Validation of diffraction data

• Use analyse.matrix anomalous for an overiew of all average anomalous differences and dispersive difference. Use analyse_anom_diff_*.distr to check individual anomalous signals in resolution shells. Use analyse_diff_*_*.distr to check individual dispersive signals in resolution shells. Use analyse_corr_*_*.dist to compare anomalous differences.
• Visual inspection of the Harker sections of anomalous difference Patterson maps is an alternative way to check anomalous signals. The Harker sections should show a number of outstanding peaks (depending on the number of anomalous scatterers).

Heavy atom search

• Search for only ~2/3 of the total number of expected sites.
• Resolution range: 15 Å and 4 to 3 Å.
• Use heavy_search.list to identify heavy atom configurations with outstanding correlation coefficients. Trials with a correlation coefficients greater than 1.5 sigma of all trials are flagged with a plus sign.
• Average maps only if a search with one set of difference structure factors was not successful.

Heavy atom model refinement, resolution of hand ambiguity

• Two heavy atom/phase refinements are carried out in parallel, one for each hand of the heavy atom configuration.
• Use all data unless there are serious doubts about the quality of the high resolution data.
• Use the data set with the smallest dispersive signal as "Reference wavelength". In general, the choice of the reference wavelength is not critical.
• Approximate values for f' and f" are usually sufficient.
• The map for the correct choice of hand should show a clear contrast between protein regions and solvent regions. The electron density of the other map is expected to be much more evenly distributed in the unit cell. If it difficult to decide which choice of hand is the correct one, density modification should be performed and the new electron density maps inspected.

Heavy atom model refinement and completion, density modification

• Use the anomalous difference Fourier map in combination with the (more powerful) gradient map to search for additional heavy atom sites.
• If the electron density is not interpretable, review the mad_phase*.summary and density_modify*.list files for the phasing and density modification procedures, respectively. The phasing power in the low resolution shells should be greater than about two. The higher the better. The phasing power normally goes down in the high resolution shells. It is often not a good sign if the phasing power is more or less constant for all resolution shells, even if it is around two. The phasing figure of merit should be 0.5 or higher.