Strategy:  The quality of an experimental electron density map will generally improve with the number of derivatives added to the phasing calculation.  To combine the heavy atom derivative data sets F_PH1, F_PH2, F_PH3, etc. for phasing two tasks must be accomplished (1) determining the correct handedness of the heavy atom model and (2) placing the heavy atoms from each derivative on the same origin.  Both tasks can be accomplished using the cross-difference Fourier method. In this method, protein phases (a_P) are first calculated using the native data set (F_P) and structure factors and anomalous differences from only one of the heavy atom derivative data sets (F_PH1   and  D_PH1).   (These phases are refered to as SIRAS phases because they are calculated using single isomorphous replacement with anomalous scattering.) A cross-difference Fourier map is then calculated using coefficients F_P-F_PH2and phases a_P.  The highest peak on this cross-difference Fourier map should correspond to the metal position of the second derivative (PH2). However, there is a 50/50 chance that we chose the incorrect hand of the first heavy atom (i.e. the heavy atom positions used for the phasing calculation).  We can judge whether the handedness we chose is correct by comparing the height of the highest cross-difference Fourier peak with the height of the highest peak from a second cross-difference Fourier map.  The second map has the same coefficients as the first cross difference Fourier map (F_P-F_PH2) but with phases a_P obtained from using the opposite hand of the first heavy atom (i.e. -x-y-z instead of x,y,z) .  The cross difference Fourier map that produces the highest peak is the correct hand.  You may have expected that the peak heights in the two cross-difference Fourier maps would be identical because of Friedel's law. That is, the phases of the protein calculated in either of the opposite hands should differ by only a negative sign.  Hence, you would expect the cross difference Fourier peaks to be the same height in either hand.  But, as you will see from the experimental evidence this is clearly not true.  The reason that there is a small but significant difference in peak heights is due to the anomalous scattering contribution D_PH1 used in calculating the protein phases.  So the protein phases output in the two phasing runs differ by more than just a negative sign (though not by much more).

Procedures:
Calculate SIRAS (single isomorphous / anomalous scattering) phases using the Europium position and calculate the cross difference Fourier map to find the pHMBA sites all in one script.    Choose the MLPHARE button in the CCP4 GUI.  Choose cross peak maps option.  Input coordinates for the Europium site (negate all signs), then select the option to "change the space group to opposite hand before running Mlphare".  Look at the log file.  Peaks from the cross difference Fourier map will be listed in decreasing order, with the highest peak on the top.  Note the position and height of the top peak.  Now, repeat the calculation but DEselect the option to "change the space group to opposite hand before running Mlphare".  Note that the highest peak in this run is higher than in the previous run.  Therefore, the space group should NOT be changed to the opposite hand.  Note the position of the highest peak and use these coordinates as the pHMBA site to input in the COMBINED phasing calculation.  Because the coordinates for the pHMBA site were taken from a difference Fourier map, they will be on the same origin as the Eu.  Deselect the option for cross-peaks maps.  Add a second derivative to the list by pressing the button to "Add Another Derivative".  Input the heavy atom site for the pHMBA.  Look at the log file and record the phasing statistics.
 
 

MlPhaRe phasing and
cross-difference Fourier calculation (original hand).
 

 MlPhare phasing combined Eu and pCMBS derivatives.