############################################################################### # Parameter names matched in common.idb: # a # b # c # alpha # beta # gamma # anom_library # coordinate_infile # fft_memory # high_res # low_res # obs_f # obs_fom # obs_pa # obs_pb # obs_pc # obs_pd # obs_phase # obs_rms # obs_sigf # parameter_infile_\d+ # reflection_infile_\d+ # reflection_infile # sg # sigma_cut # structure_infile # output_root # atom_select # write_map # target_bins # pdb_o_format # bins ############################################################################### # vars: rot_method info: Method to be used for the rotation search, there are two basic choices. Real-space superposition of Patterson maps, solutions are scored by the product function between the 2 maps. Direct rotation of the molecule followed by structure factor calculation, solutions are scored by the correlation coefficient between the observed and calculated structure factors. The real-space Patterson method is fastest but can miss possible solutions because not all Patterson vectors are included. The direct rotation method is more accurate but takes a significant amount of CPU time because an FFT must be calculated for each rotation orientation. A more time efficient method uses the direct rotation search to perform an initial coarse grid search followed by finer searches around the top peaks from the initial search. This method relies on the observation that the solution space for the rotation search is fairly smooth. # vars: patterson_low info: The minimum Patterson vector length in Å to include in the real-space Patterson method. This can be used to remove the short vectors around the origin of the Patterson map. A minimum of 5Å is recommended. # vars: patterson_high info: The maximum Patterson vector length in Å to include in the real-space Patterson method. This can be used to remove vectors which arise as a result of crystallographic symmetry. In general it is best to limit vectors to be the radius of the search model (assuming an approximately spherical shape). Highly asymmetric search models (for example, a long rod) may require a shorter maximum vector length to avoid the inclusion of too many intermolecular vectors. # vars: npeaks info: The maximum number of Patterson peaks to be used in the real-space rotation search. The Patterson peaks in the experimental Patterson map are sorted by peak height. The specified number of top peaks are taken and used in the calculation of the product function between the maps. # vars: auto_asu info: Flag determining whether the extent of the rotation search will be determined automatically based on the crystallographic symmetry. For each Laue class there is a tabulated asymmetric unit for the rotation search (see CNS_XTALLIB:cross-rf-asu.lib). Searching for possible rotation solutions in the given asymmetric unit will produce complete coverage of rotation space. # vars: tmmin tmmax t2min t2max tpmin tpmax info: User defined limits for the rotation search expressed in pseudo-orthogonal Lattman angles. These can be used to search around a subset of the full rotation space, for example limiting the search to a rotation about an axis. # vars: angle_grid info: Angular increment for the rotation search. The coarseness of the grid will have an impact on finding the correct solution but also on the CPU time taken. For the direct rotation search CPU time can be reduced by using a larger increment. The increment is usually calculated automatically by the program using:
  increment = 2 ArcSin[ high resol / (2(a+b+c)/3) ]
# vars: nlist info: The number of top rotation solution peaks to be subject to cluster analysis. The list of possible rotation solutions is reduced in size by cluster analysis. Peaks that are close together in rotation space are considered to be one peak. Rotation solution peaks are considered to be in the same cluster if they can be transformed into one another by a rotation less than a defined tolerance. This tolerance is typically defined to be 10 degrees. This cluster analysis removes grid points which are close to other grid points with higher rotation function values (product function or correlation function). # vars: fastd_gridfac info: A scale factor which determines the coarseness of the initial rotation search in the fastdirect method. The angular increment (grid size) is multiplied by this scale factor to calculate the angular increment for the coarse grid search. Increasing the increment by a factor of 4 or 5 produces a coarse enough grid to significantly reduce CPU time while still searching enough rotations to find possible solutions. # vars: fastd_npeaks info: The number of highest peaks from the coarse grid search which will be further analysed in the fastdirect method. Rotation searches using the fine angular increment are performed around the selected number of top peaks. The peaks from these searches are combined and then subject to cluster analysis. Searching the top 20 coarse rotation search peaks is a good compromise between CPU time and finding the correct solution. In difficult cases the correct solution may be found by increasing the number peaks to be searched. # vars: cluster_threshold info: Threshold in degrees for peak cluster analysis. Peaks belong to the same cluster if they are related by a rotation which is less than the threshold. A value of 10 degrees in suggested. # vars: rf_list_outfile info: Output file containing the list of rotation solution peaks. The listing file has the following format:
! index, theta1, theta2, theta3, RF-function (EPSIlon= 0.25)
     1    64.427  78.712  76.969    0.0799
     5    64.452  56.212  47.265    0.0631
     7    71.036  63.737  31.274    0.0554
The peaks are listed in Euler angles using the following convention (where all rotations are anticlockwise looking along the direction of the rotation axis): The rotation function value is the product function (real-space Patterson method) or the correlation coefficient between squared normalized structure factors (direct search). # vars: psimin psimax phimin phimax kappamin kappamax info: The range of rotation angles to be searched defined in spherical polar angluar space (where all rotations are anticlockwise looking along the direction of the rotation axis: The maximum ranges for these angles are: When looking for rotation axes (for example, a three-fold rotation axis), psi and phi should range across all values (both from 0 to 180) while the rotation angle kappa should be limited from 110 to 120 (in order to locate rotations close to 120 degrees). # vars: matrix_outfile info: Output 3d-matrix file in Mathematica format. This contains the complete rotation function as a 3-dimensional matrix. The output matrix is written in order phi, kappa, psi. # vars: atom_placed info: Atom selection which defines those atoms in the input model whose translational component is already known. These atoms will remain fixed during the translation search, but they will contribute to the structure factor calculations. The position of these atoms maybe be optimized during the final rigid body refinement if they are selected as groups for refinement. When searching for multiple copies of a molecule in the asymmetric unit it is best generate input coordinates which have a unique chain or segment identifier for each molecules (for example, A, B C etc.). If molecules A and B have already been located they can be easily selected as such with:
  (segid A or segid B)
# vars: pc_target info: Crystallographic target function to be used for Patterson correlation refinement (Brunger A.T., Acta Cryst. A46, 46-57, 1990) of molecular orientations. # vars: pc_mini_steps info: Number of steps of conjugate gradient energy minimization Patterson correlation refinement (Brunger A.T., Acta Cryst. A46, 46-57, 1990). In general 20 to 30 steps are sufficient. However, if the model is some distance from the correct orientation or there are many rigid groups more steps may be required for convergence. The translation search is generally quite sensitive to the molecular orientation so convergence of the PC refinement is desired. # vars: pcgroup_\d+ info: Atom selection defining a group of atoms that will be refined as a rigid body during Patterson correlation refinement (Brunger A.T., Acta Cryst. A46, 46-57, 1990). Atom selections for rigd groups must be disjoint. Rigid groups are usually molecules or domains. # vars: do_trans info: Flag determing whether translation searches will be performed. If this is set to false then only Patterson correlation refinement is performed. This option can be used to optimize the orientation of a list of rotation peaks without trying to determine the translation parameters. The default is to perform the translation search for each peak. The fast translation search is usually faster than the PC refinement. # vars: trans_method info: Select which translation search method is to be used. The general method usually uses the fast translation search. A phased translation search (R.J. Read and A.J.J. Schierbeek, Appl. Cryst. 21, 490-495, 1988) can be carried out if experimental phase information is available. These phases can be given in the form of Hendrickson-Lattman coefficients or centroid phases and figures-of-merit. The phased translation search algorithm cannot be used with a fixed substructure (the whole input molecule is translated). # vars: general_target info: Defines the target for the general translation search method. Although a variety of targets are possible, in order to use the fast translation search algorithm the fastf2f2 target should be used. Other targets use a conventional translation search which is between 200 and 500 times slower than the fast translation search. # vars: optimize_sum info: Determines the memory usage of the summation stage of the translation search. Both the fast and the direct translation search can utilize less memory. However, there is a cost associated with the use of less memory. For the fast translation search if there are many symmetry operators (more than 24) then using less memory will result in a small increase in the CPU time for the search. In the fast translation search using less memory with fewer than 24 symmetry operators saves both memory and decreases CPU time. For the direct translation search using less memory significantly increases the CPU time so should be avoided unless really necessary. By default the automatic option should be used for this parameter, this will select the appropriate memory usage depending in the translation search method and spacegroup symmetry. The default can be overridden in the case of limited memory availabilty. For example with a high symmetry spacegroup (such as F-centered cubic) by default the fast translation search will use more memory in order to reduce CPU time. If this memory in not physically available then the job may terminate. In such an instance it is appropriate to change this option to "memory" to force the fast translation search to use less memory (but take more CPU time). # vars: rf_list_infile info: Input file containing the list of rotation solution peaks. The listing file has the following format:
! index, theta1, theta2, theta3, RF-function (EPSIlon= 0.25)
     1    64.427  78.712  76.969    0.0799
     5    64.452  56.212  47.265    0.0631
     7    71.036  63.737  31.274    0.0554
The peaks are listed in Euler angles using the following convention (where all rotations are anticlockwise looking along the direction of the rotation axis): The rotation function value is ignored. # vars: max_rf_peaks info: The maximum number of peaks from the input rotation list file to be analysed. This must be less than or equal to the number of peaks in the file. The fast translation search is very fast, therefore it is practical to test many of the rotation peaks (even 100 or 200 peaks can be analysed). # vars: grid_int info: Extent in degrees of a fine rotation search around each input rotation peak. This can be used to brute-force search possible orientations close to the starting orientation. This is not normally necessary as the optimization in carried out by Patterson correlation refinement. # vars: delta info: Angular increment for fine rotation search around each input rotation peak. # vars: tf_grid info: Resolution of the translation search grid as a fraction of the highest resolution limit in reciprocal space (e.g. of the diffraction data). For example, given tf_grid=0.33, if the high resolution limit of the diffraction data is 4Å, the grid spacing of the map is 1.32Å in each dimension. The grid should be 0.33 or smaller. # vars: ntrials info: The number of translation search peaks to be tested for each rotation peaks. The translation search generates a list of translation search peaks. Usually the highest peak is taken and subjected to Patterson correlation refinement. With this parameter it is possible to take more than just the first translation peak. The specified number of translation peak solutions for this rotation peak are subject to Patterson correlation refinement and then ranked by the final correlation coefficient. This option may be useful in difficult cases but in general the highest peak from the translation search is the correct one (if the rotation is close to correct). # vars: packing_anal info: Perform packing analysis on the translation solutions. A mask is calculated around the final model and the percentage of the unit cell volume covered by this mask is calculated. Translation solutions which are non-overlapping will have higher packing values. # vars: rigid_target info: Crystallographic target function to be used for final rigid body refinement of molecular orientations. # vars: final_mini_steps info: Number of steps of conjugate gradient energy minimization rigid body refinement. In general 20 to 30 steps are sufficient. # vars: rigidgroup_\d+ info: Atom selection defining a group of atoms that will be refined as a rigid body . Atom selections for rigid groups must be disjoint. Rigid groups are usually molecules or domains. # vars: write_coord info: Write the coordinates of each translation solution as unique PDB files. This can be useful for visual analysis of the solutions. # vars: trans_list info: Write the translation solution peaks for each rotation peak to unique files. A maximum of 100 peaks are listed for each translation search. # vars: sol_auto info: automatic bulk solvent parameter search # vars: sol_output info: optional file with a listing of the results of the automatic bulk solvent grid search # vars: sol_rad info: solvent mask parameter # vars: sol_shrink info: solvent mask parameter #