3. Summary of operations
This section details flowcharts for the various operations required for structure solution, using as examples the programs for MR in the CCP4 package, assuming that a merged reflection data file (in "MTZ" format) and a coordinate file (in "PDB" format) are available. In fact two alternative strategies are presented, the only essential difference being that the first uses F's (possibly "sharpened"), and the second uses E's (normalised amplitudes). The first is more of a "black box" procedure in that it uses only the AMORE program, and it will no doubt easily solve the majority of structures. The second is more complicated in that some auxiliary programs are needed; however it will be demonstrated that it succeeds in some difficult cases (such as NCS) where the "black box" procedure has difficulty.
3.1. AMoRe rotation function using F's
3.1.1. AMoRe SORTFUN
Convert the reflection data file into the internal format used by the AMORE program (AMORE / SORTFUN).
3.1.2. AMoRe TABFUN
Optionally rotate the search model so that its principal axes of inertia are parallel to the coordinate axes, and the centroid of the coordinates is shifted to the origin. This helps to minimise the size of cell required in the next step. Also compute a molecular Fourier transform for use in structure factor calculation by interpolation (AMORE / TABFUN).
3.1.3. AMoRe ROTFUN GENERATE
Compute structure factor amplitudes for the search model in space group P1 in a large rectangular cell (AMORE / ROTFUN / GENERATE).
3.1.4. AMoRe ROTFUN CLMN
Compute spherical harmonic coefficients for the target and model Pattersons, optionally using "sharpened" F's (AMORE / ROTFUN / CLMN).
3.1.5. AMoRe ROTFUN ROTATE
Think about the expected asymmetric unit of the cross-RF, and compute it as a function of the Eulerian angles (a, b, g). There should be a peak for each molecule in the asymmetric unit (AMORE / ROTFUN / ROTATE).
3.2. Rotation function using E's
Modify the PDB header in the coordinate file, so that it has CRYST and SCALE records for the chosen cell in the RF (PDBSET).
Compute structure factor amplitudes in space group P1 for the search model in a large rectangular cell (SFALL).
Normalise both the observed and calculated amplitudes (ECALC - 2 runs).
3.2.4. AMoRe SORTFUN
Convert the normalised observed and calculated reflection data files into the internal format used by the AMORE program (AMORE / SORTFUN - 2 runs).
3.2.5. AMoRe ROTFUN CLMN
Compute spherical harmonic coefficients for the target and model Pattersons, (AMORE / ROTFUN / CLMN).
3.2.6. AMoRe ROTFUN ROTATE
Think about the expected asymmetric unit of the cross-RF, and compute it as a function of the Eulerian angles (a, b, g). There should be a peak for each molecule in the asymmetric unit, usually the highest (AMORE / ROTFUN / ROTATE).
In each case, the low and high resolution cutoffs and the radius of integration should be varied, and the peak lists examined for consistency. If possible also check for consistency between different models (rotate first into a common orientation).
3.2.7. POLARRFN & PLTDEV, then AMoRe ROTFUN ROTATE RFCORR
In the case of NCS, using E's compute and plot the self-Rotation function in terms of spherical polar angles (POLARRFN & PLTDEV). Then recompute the self-Rotation function in terms of Euler angles and check for consistency with the cross-RF peak list (AMORE / ROTFUN / ROTATE, RFCORR).
3.3. AMoRe translation function using F's
3.3.1. AMoRe TRAFUN
For the best 10-20 RF solutions, compute the TF; in case of difficulty, try varying the resolution cutoffs and sharpening factors (AMORE / TRAFUN).
3.3.2. AMoRe TRAFUN 2
In the case of NCS, fix the first molecule found and compute the TF for the second; proceed stepwise in the same manner for the remaining molecules (AMORE / TRAFUN).
3.3.3. AMoRe FITFUN
For the best 10-20 TF solutions, do rigid-body refinements and choose the solution with the highest correlation coefficient (AMORE / FITFUN).
3.3.4. AMoRe TABFUN / PDBSET
Apply the rotation matrix and translation vector to the model coordinates output by AMORE / TABFUN (PDBSET).
3.4. Translation function using E's
For each molecule in the asymmetric unit apply the appropriate rotation matrix, calculated from the Eulerian angles of the peak(s) in the RF, to the model coordinates. Also modify the PDB header in the coordinate file, so that it has CRYST and SCALE records for the target cell (PDBSET - run for each molecule in the asymmetric unit).
Calculate structure factor amplitudes and phases (SFALL - run for each molecule in the asymmetric unit).
Combine all the columns for the observed reflection data with those for the calculated molecules into a single file (CAD).
Normalise all the columns of amplitudes and compute the Fourier coefficients for the intra-molecular vector subtracted full-symmetry Translation function (TFFC).
Think about the asymmetric unit of the TF. This is usually a fraction of the unit cell, for example in polar space groups the TF map is 2-dimensional. Compute the Fourier transform to get the Translation function map (FFT).
Search the Translation function map for peaks; one peak should stand out from the rest with relative signal/noise > 1 (MAPSIG).
3.4.7. TFFC, then FFT and MAPSIG
In cases of non-crystallographic symmetry it is necessary to place all molecules relative to the same origin with non-crystallographic Translation functions; note that the asymmetric unit is always a whole primitive cell (TFFC, then FFT and MAPSIG as before for each molecule in the asymmetric unit).
Apply this translation vector to the rotated model coordinates (PDBSET).
3.5. Computer graphics
Finally, view the transformed molecules in the unit cell on the computer graphics and check for good packing.