2.10                      *******************
                                  * MRGBDF WRITE-UP *
                                  *******************
        
           MRGBDF is a program to create coefficients for Bijvoet difference 
        or cross Bijvoet difference Fourier synthesis calculations, i.e. try 
        to determine anomalous scatterer locations from phase information 
        obtained from one or more other derivatives. It can also be used 
        simply to search for additional anomalous scatterer sites once initial 
        estimates of protein phases become availabler, although the
        "difference coefficients" file output from PHASIT may be better suited
        in this case since one can then also subtract out the heavy atoms
        already present in the model, and also generate a "calculated"
        Patterson for comparison with the "observed" one. The program is 
        interactive and prompts for the names of the input and output files, 
        and a d spacing cutoff. A merged file input may be either a "native 
        anomalous scattering" or "derivative anomalous scattering" type file 
        (see PHASIT write-up), and the user will be prompted to identify the 
        type. The output file can be used in FSFOUR with MAPTYP=8 to compute 
        the Bijvoet-difference Fourier synthesis.  If the input Bijvoet pair 
        file is one of the "merged" data files originally input to PHASIT, 
        then the coefficients output can be used to compute a Bijvoet 
        difference Fourier to identify additional anomalous scatterer sites 
        which may have been overlooked. In that case it is not a "cross" 
        Fourier but a straight Bijvoet difference Fourier. If the Bijvoet pair 
        file input corresponds to a new derivative, then the coefficients can 
        be used for a "cross" Bijvoet difference Fourier to reveal the 
        locations of anomalous scatterers in the new derivative. The program 
        is also useful to determine whether the heavy atom hand designation
        agrees with the anomalous scattering data. When everything 
        is consistent maps computed should reveal POSITIVE peaks at the
        appropriate anomalous scatterering sites for all Bijvoet pair data
        sets. If the hand for the heavy atoms used in phasing is inconsistant,
        then the map will reveal NEGATIVE peaks at the true sites, i.e. those
        related to the input (incorrect) set BY A CENTRE OF SYMMETRY.
        Program PSRCH can be used to list the strongest peaks (both positive
        and negative) in the map, and program HNDCHK can be used to aid in
        hand determination by examining the density precisely at any arbitrary
        location. In general, one could phase the data using both possible
        hands and check the results as just described. 
           If derivative Bijvoet differences are used, The program will prompt
        the user to supply values for derivative to native scale and delta B
        factors, if rescaling is requested. If utilized, this option enables
        the user to change the scaling originally carried out in CMBANO, to
        reflect the fact that additional scattering power is present in the
        derivative data set. In that case the new scaling parameters should be
        those determined from PHASIT in phase refinement mode.                          
        
                                   ***** FILES *****
        
        The input Bijvoet pair file is identical (in form) to one of the 
        "merged" data files input to PHASIT. Each record should contain either
        
        H, K, L, F+, SIG(F+), F-, SIG(F-)
        
                      or
        
        H, K, L, FP, SIG(FP), FPH+, SIG(FPH+), FPH-, SIG(FPH-)
        
        This file refers to the new derivative which is to be solved, or to a 
        current data set for which one wants to search for additional 
        anomalous scatterer sites. It is read in free format.
        
        The input protein phase file can be one of two types. Usuallly it will 
        be the last output file from BNDRY, or an output file from PHASIT (in 
        protein phasing mode). In general, it should contain the best 
        available phases. The form of the file would then be identical to that 
        output from BNDRY or PHASIT (in protein phasing mode). It is also 
        possible however, to input a protein phase file which contains records 
        with the "short" reflection file form (only h,k,l,fo,fc,phi) as 
        generated by GREF or PHASIT (in structure factor calculation mode, 
        IHLCF=0). In that case there is no figure of merit present, thus FOM= 
        1. is used during generation of the output coefficients. This would be 
        the case if the protein phases come from a complete (or partial) 
        protein model based structure factor calculation. The program can 
        automatically determine which type of file was input. Note however, 
        that a phase file generated by GREF should NOT be used here unless
        GREF was used to compute structure factors from a complete protein
        model.
         
        The output file is binary and is suitable for input to FSFOUR.
        Each record contains
                                                     
        H, K, L, FOM*F+, FOM*F-, PHI+
        
        where the indices are INTEGERS, the other quantities REALS and PHI+ is
        in degrees.  
        
        PHI+  =  PHICALC     for + type output file.
        
        PHI+  = -PHICALC     for - type output file.
        
        
        The figure of merit and PHICALC come from the phased file while
        F+ and F- come from the Bijvoet pair data file. The Bijvoet difference
        Fourier should be computed with coefficients
        
          -i * (FOM*F+  - FOM*F-) * exp (i * PHI+)
        
        where the -i factor is applied after expansion to a hemisphere, and 
        during the expansion, care is taken to insure that the differences are 
        "flipped" if putting the reflection into the desired hemisphere 
        involves an inversion. This is taken care of automatically in FSFOUR 
        if MAPTYP=8 is selected.