############################################################################### # Parameter names matched in common.idb: # a # b # c # alpha # beta # gamma # fft_memory # grid # high_res # low_res # reflection_infile_\d+ # sg # output_root # ncs_infile # map_format # reset_b # reset_q # grid_res # sitedatabase_infile(\.\d+)? sitedatabase_outfile(\.\d+)? ############################################################################### # vars: pc_target info: Target function for Patterson Correlation Refinement (Brunger (1990), Acta Cryst. A46, 46-57). This can be the correlation coefficient between the observed squared structure factors and the calculated squared structure factors (F2F2), or the correlation coefficient between the observed normalized squared structure factors and the calculated normalized squared structure factors (E2E2). In theory, correlation coefficients can be in the range between -1 (perfectly anti-correlated) and 1 (perfectly correlated). In practice, correlation coefficients range from values close to zero (very poor correlation of observed and calculated structure factors) to about 0.6 or 0.7 (very high correlation). # vars: nstep info: Number of Patterson Correlation Refinement steps for the refinement of atomic coordinates. The refinement will not converge if this value is too low, and take very long if the value is too high. # vars: bnstep info: Number of Patterson Correlation Refinement steps for the refinement of B-factors and occupancies. The refinement will not converge if this value is too low, and take very long if the value is too high. # vars: shift_dampen info: Flag for Patterson Correlation Refinement. Turns shift dampening on or off. To avoid numerical instabilities, shift dampening should normally be turned on. # vars: ref_occupancy info: Flag for Patterson Correlation Refinement. A negative value turns the refinement of occupancies off. Otherwise the occupancies are refined if the number of heavy atoms placed is greater than or equal to the given number. The refinement of occupancies is useful mainly for isomorphous replacement data. # vars: ref_bfactor info: Flag for Patterson Correlation Refinement. A negative value turns the refinement of B-factors off. Otherwise the B-factors are refined if the number of heavy atoms placed is greater than or equal to the given number. For structures with many heavy atoms sites (e.g. SeMet substituted proteins) it can be useful to set ref_bfactor to about 1/4 or 1/3 of the number of expected sites. # vars: new_sites info: Number of heavy atoms sites to be found. In most cases it is best to search only for about 2/3 of the number of expected sites. The missing sites with low occupancies or high B-factors can then be found with difference Fourier techniques using experimental phases obtained with the task files ir_phase.inp or mad_phase.inp. # vars: site_type info: Chemical type of the heavy atoms sites. If gaussian form factors are used, this type is used to determine the form factors from the CNS scattering library file. # vars: scatter_mode info: Possible values for scatter_mode are "gaussian" or "constant". The gaussian form factors (which depend on the diffraction angle of a given reflection) are taken from a CNS scattering library file. This is the appropriate choice for isomorphous replacement difference data. In constant mode, a unit form factor is used for all reflections. This is the appropriate choice for anomalous difference data. # vars: scatter_library info: Path name for the CNS scattering library file containing atomic (gaussian) form factors (scattering factors) for all chemical elements and some ions. # vars: init_bfac info: The initial B-factor assigned to newly placed heavy atom sites. In most cases, an initial value of 20 is appropriate. The B-factors are normally adjusted by Patterson Correlation Refinement. # vars: number_of_maps info: The number of Patterson maps to be averaged. Two or more sets of difference structure factors can be averaged and optionally weighted by diffraction ratios (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). However, in most cases it is possible to use only the diffraction data for the wavelength with the largest anomalous or isomorphous signal to locate the heavy atom sites. # vars: weight info: Flag for the averaging of two or more Patterson maps. If set to true, a statistical weighting is carried out by grouping the reflections into resolution shells (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). # vars: bins_w info: Number of resolution shells for statistical weight calculation, removal of Patterson origin peaks, and Patterson sharpening by E-normalization. All these operations are carried out for groups of reflections in resolution shells of equal volume. # vars: bins_w_hs info: Number of resolution shells for statistical weight calculation. Two or more Patterson maps can be averaged. A statistical weighting is carried out by grouping the reflections into resolution shells (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). # vars: map.\d+.mode info: Type of Patterson map. Possible types are "native" (used with pre-processed FA structure factors), "isomorphous" (isomorphous replacement), "dispersive" (MAD), and "anomalous" (SAD or MAD). # vars: map.\d+.[fs]_[ab] info: Name of reciprocal space array with structure factors or estimated-standard-deviations. The array must be defined in a reflection file. # vars: map.\d+.cut_f info: Reflections with a structure factor smaller than cut_f times the estimated-standard-deviation are rejected (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). # vars: map.\d+.cut_df info: Reflections with isomorphous, dispersive or anomalous differences smaller than cut_df times the estimated-standard-deviation are rejected (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). # vars: map.\d+.max_df info: Root-mean-square outlier cutoff as defined by Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55. Reflections with |delta F| > max_df * rms(|delta F|) will be rejected. This means, reflections with very large isomorphous, dispersive or anomalous differences are rejected. max_dfis usually set to a value between 3.5 and 5. # vars: map.\d+.kscale map.\d+.bscale map.\d+.nscale_iter info: Overall k-scaling and B-scaling is used to compensate for systematic errors caused by differences between crystals and data collection conditions. The scaling and the B-factors are obtained by least-squares minimization (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). kscale is normally set to "yes", and bscale is set to "anisotropic". The number of scaling iterations is given by nscale_iter. Four cycles are appropriate in most cases. A summary of the overall k-scaling and B-scaling can be found at the top of the summary file (e.g. heavy_search.summary). # vars: ntrials info: Number of trials. The heavy atom search procedure consists of four stages. In the first stage, the observed diffraction intensities are filtered by various cutoff criteria and then used to compute native structure factors or difference structure factors. The second stage consists of a Patterson search. A given number (ntrials, typically 100) of highest peaks in the resulting Patterson search map are sorted and subsequently used as initial trial sites. The third stage consists of a sequence of alternating Patterson searches and Patterson Correlation refinements starting with each of the initial trial sites. This stage produces a large a number of potential solutions. The final stage consists of sorting these solutions ranked by the value of the target function of the Patterson Correlation refinement. If the correct solution is found, it is normally characterized by the highest value of the target function and a significant separation from incorrect solutions (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). # vars: first_trial info: Peak number of the first Patterson search that will be used for the first trial. ntrials peaks will be tested after this peak. first_trial is useful to re-start a heavy atom search which crashed. # vars: search_method info: Selects the Patterson search method. "reciprocal space" selects the Translation Function, "direct space" selects the Symmetry Minimum Function for the first Patterson search and the Image Seeking Minimum Function for the subsequent searches. "reciprocal*direct" combines the results of both the Translation Function and the Minimum Functions. In general, the "reciprocal*direct" search method is the most powerful method because it finds more correct or partially correct solutions and they appear sooner during the search (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). # vars: w_imf info: Dynamic weight factor for reciprocal*direct Patterson search method. The Image Seeking Minimum Function (IMF) can be viewed as a likelihood map that provides the likelihood of each grid point to be close to an additional heavy atom site, and ideally, Fast Translation Function (FTF) and IMF maps would be identical. However, the correlation coefficient between both maps is typically only in the range of 0.1 to 0.4 and decreases with the number of existing sites. This is not only due to noise, systematic errors resulting from the use of difference data, and the different methods of generating the likelihood maps, but also reflects a shortcoming of the IMF the FTF. The IMF algorithm assumes that all Patterson peaks are resolved. Peaks which coincidentally overlap are incorrectly weighted. Therefore a combination of FTF and IMF is only useful for searches for the first few additional sites, when Patterson peaks of previously placed sites are not likely to be overlapped. The decreasing accuracy of the IMF is empirically taken into account by the weighting factor w_imf. In the search for the first additional site, the IMF is normalized by linear scaling and shifting such that the minimum value is zero and the maximum value is one, and each grid point of the IMF is then used as a multiplicative weight for the FTF map. In the search for the second additional site, the IMF is scaled and shifted such that the minimum value is 1/w_imf, for the third site it is 2/w_imf, and so on. The maximum value is always one. If the minimum value is close to one, the IMF is no longer used. Typically, w_imf is set to 3 (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). # vars: special info: Flag for special positions. If false (the default), heavy atoms are not placed on special positions (e.g. a rotation axis). If true, all positions are allowed. # vars: mate_low_cut info: Lower distance cutoff for symmetry mates. Positions with symmetry mates closer than the given distance are excluded. Typically, a distance cutoff in the range from 3.5Å to 6Å is used. # vars: low_cut info: Lower distance cutoff for atoms which are not related by symmetry. In the heavy atom search algorithm, positions which are closer than the given distance to any of the previously placed atoms are excluded. Typically, a distance cutoff in the range from 3.5Å to 6Å is used. # vars: expected_corr_inc max_no_corr_inc info: The correlation coefficient before placing a new atom is compared to the correlation coefficient after the addition and refinement of the new atom. If the correlation coefficient did not increase by the amount given by expected_corr_inc (typically 0.01), the search for that particular initial trial site is deemed to have reached a dead end and no additional atoms are placed. However, experience shows that for structures with many heavy atom sites (> 20) the correlation coefficient sometimes does not significantly increase when adding a correct new atom. Therefore, the parameter max_no_corr_inc was introduced. It specifies the total number of times a decrease or no change in the correlation coefficient is tolerated upon addition of a new atom. The main drawback of increasing max_no_corr_inc is that the total run-time for the heavy atom search increases as more time is spent on incorrect solutions. # vars: lessmemory info: Flag for the selection of the implementation of Fast Translation Function (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55). Given lessmemory=false, three-dimensional arrays are used. Given lessmemory=true (the default), two-dimensional arrays are used. The latter choice does not only require significantly less memory, but is also faster for most structures. The exception are large high-symmetry (hexagonal or cubic) structures. # vars: export_f_patt info: File name for F_patt structure factors (Grosse-Kunstleve & Brunger (1999), Acta Cyrst. D55) used in the heavy atom search. Optionally these structure factors can be exported for use in an external program. The structure factors are written in the standard CNS format. No file will be written if the file name is blank. # vars: sharpen info: Flag for sharpening of Patterson map. If true, normalized structure factors (E-values) are used for the computation of the Patterson map. # vars: origin_remove info: Flag for removal of the origin peak in a Patterson map. If true, the origin peak is removed by substracting the statistical average of all structure factors from each individual structure factor. # vars: harker_sections info: Flag to request a plot of the Harker sections. If true, a map file is written for each Harker section which is parallel to one of the faces of the unit cell. Due to a limitation of the program, Harker sections which are not parallel to one of the unit cell faces cannot be plotted. The plot_patterson utility program can be used to contour the map files with the Harker sections and generate Postscript files for printing. # vars: plot_unitcell info: Flag for extent of the Patterson map and the Harker sections. If false, only the grid points in a minimal box or rectangle, respectively (with edges parallel to the unit cell edges), containing an asymmetric unit are written to the map files. If true, the Patterson maps cover the whole unit cell, and the Harker sections cover the corresponding whole face of the unit cell. # vars: 3d_patterson info: Flag for output of 3-dimensional Patterson map. If true, the Patterson map is written to a map file. # vars: peak_list_sigma peak_pdb info: peak_list_sigma defines the sigma cutoff for the generation of a list of Patterson peaks. Only peaks greater than the specified sigma value will be listed. The peak list is written to a list file, for example patterson_map.list. A negative value disables the peak search.

peak_pdb is a flag for the generation of a PDB file with the list of Patterson peaks. If true, the peaks shown in the list file will also be written out in PDB format (for example patterson_map.pdb). # vars: show_info info: Select amount of information shown in Harker sections. Choices are "none", "short", and "all". If all" is selected, the information shown can sometime clutter the plots of the Harker sections. In such a case, "short" or "none" should be selected. # vars: macro_cycle info: Number of macrocycles. One macrocycle consists of Patterson Correlation refinement of the atomic coordinates, followed by B-factor and occupancy refinement. # vars: occupancy info: Flag for occupancy refinement. Only if true, occupancies are refined by maximizing the Patterson correlation. # vars: bfactor info: Flag for B-factor refinement. Only if true, B-factors are refined by maximizing the Patterson correlation. # vars: write_fcalc info: Write structure factors calculated from the atomic model to a hkl file (for example predict_patterson.hkl). The structure factors are written in the standard CNS format. 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 #