Google

AMORE (CCP4: Supported Program)

NAME

amore - Jorge Navaza's state-of-the-art molecular replacement package, updated February 1999.
The SORTFUN and TABFUN output are NOT compatible with the old version.
New keyword CRYSTAL for TABFUN.
[Keyworded input]

CONTENTS

  1. DESCRIPTION
  2. Further details of the PROGRAM FUNCTIONS
  3. LIKELY PROBLEMS
  4. KEYWORDED INPUT
    1. SORTFUN keywords
    2. TABFUN keywords
    3. ROTFUN Keywords
    4. TRAFUN keywords
    5. FITFUN keywords
    6. REORIENTATE keywords
  5. NOTES
    1. Memory allocation
    2. Rotation matrix definitions
    3. Orthogonalisation codes
  6. EXAMPLES
  7. AUTHORS
  8. REFERENCES
  9. SEE ALSO

DESCRIPTION

AMoRe includes routines to run a complete molecular replacement.

As well as carrying out ROTATION and TRANSLATION searches against various targets, and doing RIGID BODY REFINEMENT, there are routines to reformat the observed data from the new crystal form, and to generate and tabulate structure factors from the model in a large P1 cell. See reference [1].

The steps are usually carried out in the following order.

  1. The observed data is extended to cover a hemisphere of reciprocal space and reformatted.
  2. Structure factors for the model are tabulated on a fine grid (corresponding to a large "unit cell"). This is the key to the program's speed. All subsequent structure factors required for the searches are obtained by interpolating into this table. The structure factors can be calculated within Amore from a set of coordinates, using the option TABFUN, or generated outside the program and read in using the option SORTFUN.
  3. The rotation function is run searching for Patterson correlation within a sphere centred on the origin. This allows the Patterson to be expressed in terms of spherical harmonics, and the calculation to exploit FFT techniques. Two different types of indicators of a good solution are given (see also below)
    a) The correlation between the observed and model pattersons;
    b) Correlation coefficients and Rfactors between the observed Fs or Is and generated Fs or Is from a model with the given orientation.

AMORE requires a LOT of memory and this may cause problems on some machines. However this new release is considerably less demanding than the older one. (see Memory allocation).

Further details of the PROGRAM FUNCTIONS

Step_1 SORTFUN - reading, extending, sorting and reformatting a list of reflections

Input: HKLIN
Standard MTZ file (maybe observed data, structure factors generated by some technique, E values from ECALC, etc.)
Input: Memory allocation parameters
SORTING_NR
(default 100000)
Output: Either HKLPCK0 (see option 1)
Packed file of H K L F [SIGF or PHI] NMULT in hemisphere: h,+-k,+-l. This is a binary file which also holds the unit cell, symmetry operators, and maximum h, k,l and resolution (see example [1]a).
Output: or TABLE<i> (see option 2)
Table of the finely sampled inverse Fourier coefficients (i.e., structure factors which have been read in from a previously prepared MTZ file). These must extend a little past the required resolution of the calculations to allow for interpolation. This is a binary file which also holds the large "unit cell", maximum h, k, l, and resolution (see example [1]b).

Option 1:

  • Packs and sorts H K L Fobs [SigFobs or PhiObs] to an internal form for use in later steps.

Option 2:

  • Packs an input list of H K L FC PHIC for use as a TABLE. This format is described below. This gives the user great flexibity to try different types of search models. For example, structure factors can be generated from modified electron density maps, or calculated structure factors can be converted to E values (see example [1b]).

Step_2 TABFUN - reading model coordinates, repositioning them and generating structure factors from them

Input: XYZIN<i>
Standard PDB file for the model
Input: Memory allocation parameters
TABLING_MI, TABLING_MR, TABLING_MC
(defaults 5000000, 4400000, 100000)
Output: XYZOUT<i>
Coordinates after repositioning
Output: TABLE<i>
Table of the finely sampled structure factors generated from the shifted model, and calculated for a large "unit cell".
Output: Log File
Contains vital information about the coordinates which will be used at later stages of the procedure (e.g. Minimal Box, Centre of Mass, Rotation, Maximal distance from Centre of Mass)
Optional Output: HKLOUT<i>
This is rarely used, but can be useful for checking purposes. ASCII file of finely sampled inverse Fourier coefficients as H K L FC PHIC (i.e. structure factors)

The procedure is:

  1. The model coordinates are translated so that their centre of gravity is at the origin. They can then be rotated so that the principal axes of inertia of the model are parallel to the a, b and c axes of the "minimal box" which just contains the model. The dimensions of the "minimal box" are determined, and the "maximum distance" of any coordinate from the centre of mass.
    You may choose not to ROTATE the model; in some case results may then be simpler to interpret. For instance if you want to compare results from several models it is convenient to allow the first model to ROTATE, then to fit all others to these repositioned coordinates which will have benn output to the assigned XYZOUT. It may also be useful if you expect some predictable result; e.g. that the new crystallographic symmetry axes should map onto those of the model structure.
    Hint: It can help to understand results if some "pseudo" atoms are added to the model PDB file. For example if you have a two fold axis in the original structure put 2 coordinates on this axis. If the model forms a tetramer centred at (Xt,Yt,Zt) include this coordinate plus 3 which lie on the tetramer axes.
  2. Structure factors are generated from the modified coordinates for a "CELL" with dimensions SCALE*minimal_a, SCALE*minimal_b, SCALE*minimal_c and all angles = 90 . SCALE has the default value of 4, but can be reset by the SAMPLE keyword. All later structure factors and gradients for the model in its various orientations are interpolated from this data. 
       Expected Error in R factor with SCALE = 4  -  3 %
   Expected Error in R factor with SCALE = 3  -  9 %
   Expected Error in R factor with SCALE = 2  - 17 %
    You may need to generate TABLEs for several models, e.g. for different domains. Up to four different TABLE<i> files can be assigned during the translation search, and for rigid body refinement.

Step_3 ROTFUN

Runs the rotation function. Does the following four stages (they can be run seperately but I can think why..).

Step_3a GENERATE_Stage

Keyword: GENERATE - calculates structure factors for model in a suitable cell, and packs them in the same format as the output of SORTFUN.

Input: TABLE<i>
See above
Input: Memory allocation parameters
ROTING_MI, ROTING_MR, ROTING_MC, ROTING_MD
(defaults 500000, 600000, 2200000, 20000)
Output: HKLPCK1.

Step_3b CALCULATE_Spherical_HARMONICS_Stage

Keyword: CLMN - calculates spherical harmonics for crystal and models.

Input: HKLPCK<i> (HKLPCK0 for crystal, HKLPCK1 for model)
Output: CLMN<i>

Step_3c ROTATION_Stage

Keyword: ROTATE - calculates rotation function and finds many possible solutions by Patterson overlap.

Input: CLMN<i>. (CLMN0 for crystal, CLMN1 for model)
Output:
  • For the CROSS rotation function the output rotational solutions are given in terms of the Eulerian angles, alpha, beta and gamma with each line flagged: SOLUTIONRC.
    The Eulerian angles use the convention described by Tony Crowther which is used in all CCP4 programs, e.g. ALMN, LSQKAB, PDBSET and DM. They define a rotation matrix which moves the model molecule into the proper orientation for the new crystal form. The model is first rotated through gamma about Zo, then through beta about the new Yo, then through alpha about the new Zo. Positive rotation is clockwise when looking along the axis from the origin. See elsewhere for details of the definitions of the rotation matrix and the orthogonalisation conventions which define Zo Yo and Xo.

    Four solution criteria are tabulated:

    • CC_F is the correlation coefficient between the observed amplitudes for the crystal and the calculated amplitudes for the model. It is surprising that this is a satisfactory target, since the model amplitudes are generated in a P1 cell, but it does seem to be the most effective discriminator. It is sensible to sort the solutions on this target.
    • RF_F is the classic R factor between the observed amplitudes for the crystal and the calculated amplitudes for the model. Again, it is surprising that this is a satisfactory target, since the model amplitudes are generated in a P1 cell, but it does seem to be reasonable, although the CC_F is probably better.
    • CC_I is the correlation coefficient between the observed intensities for the crystal and the sum of calculated intensities for all symmetry equivalents of the model, i.e. the intensities are summed, but without any correction for the relative positioning of the symmetry related molecules.
    • CC_P is the Patterson correlation coefficient between the crystal and the model pattersons evaluated within the defined sphere centred on the Patterson origin.

  • For the SELF rotation function the solution is given in terms of Eulerian and polar angles with each line flagged: SOLUTIONRS.
    If Kappa is 180 or 120 then you may have a 2-fold or a 3 fold rotation between NCS related molecules.
    If you expect higher symmetry, e.g. 222 complex, check that the angles between related axes are perpendicular (Test DC_X1*DCX2 + DC_Y1*DCY2 +DC_Z1*DCZ2 = 0).
Output: A map of the rotation function can be output in the standard CCP4 format. This is assigned to MAPOUT and can be contoured in the usual way (NPO). It is sectioned along beta.

Step_3d REORIENTATE_Stage

Keyword: SHIFT - converts the Eulerian angle solutions determined for the model stored in XYZOUT<i> to give solutions to be applied to original MODEL.

Input: Centre of Mass and Eulerian angles which were applied to the original MODEL in TABFUN.
Output: Some rotational solutions appropriate for the original coordinates.
This can be replaced by PDBSET; see example [3]d.

Step_4 TRAFUN

Calculates the translation function using various target options.

Input: HKLPCK0
Crystal h k l output by SORTING step.
Input: TABLE<i>
For any model(s) you wish to use.
Input: Memory allocation parameters
TRAING_NR, TRAING_MEQ, TRAING_MRT, TRAING_MT, TRAING_MR
(defaults 20000, 24, 2000000, 2200000, 1000000)
Input: A list of solutions to the Rotation function output obtained in Step_3.
The search for several molecules can be done by finding first one molecule, then FIXing it whilst searching for a second molecule, etc.
Output: A list of solutions flagged as: SOLUTIONTF.
Each has: Alpha_i Beta_i Gamma_i Xf_i Yf_i Zf_i CC_F RF_F CC_I Dmin.
The Xf, Yf and Zf are fractions of the observed unit cell edges. CC_F RF_F CC_I are described above. Dmin is the shortest distance between the centres of mass of the symmetry equivalent molecules.
Output: A map of the translation function can be output in the standard CCP4 format.
This is assigned to MAPOUT and can be contoured in the usual way (NPO). The same file assignment is used for each TRANSLATION search you make, so if you want to contour your favourite solution you will need to rerun the calculation with only that SOLUTION. Remember it may be very large; assign it to a scratch area, or /dev/null if this causes problems.

Step_5 FITFUN

Performs rigid-body refinement for any specified solution of the rotation or translation search, see reference [5].

Input: HKLPCK0
Crystal h k l output by SORTING step.
Input: TABLE<i>
For any model(s) you wish to use.
Input: Memory allocation parameters
FITING_MEQ, FITING_MT, FITING_NR, FITING_NP
(defaults 24, 2200000, 20000, 6)
Input: A list of solutions.
Output: A list of solutions flagged as: SOLUTIONF.
They are given as: Alpha_i Beta_i Gamma_i Xf_i Yf_i Zf_i CC_F RF_F CC_I with the conventions described above.

Check that the CCs and RF_F have improved.

Step_6 REORIENTATE

This works out the appropriate rotation and translation parameters to apply to the initial model (can also be done while running ROTFUN or FITFUN.)

Input: Centre of Mass and Eulerian angles which were applied to the original MODEL in TABFUN.
Input: The refined rotation and translation parameters output by FITFUN.
Input: HKLPCK0
To extract the unit cell of new crystal form.
Output: A list of solutions given as:
Alpha_i Beta_i Gamma_i XA_i YA_i ZA_i Correlation_coefficient_i Rfactor_i. The XA, YA and ZA are given in Angstroms. Each line is flagged: Shifted_sol.

LIKELY PROBLEMS

Some common errors:

  • You must run both CLMN calculations with the same resolution limits and sphere radius.
  • The HKLPCK files all pack the hkl and symmetry flag into one integer. The program checks the maximum values of H K L and NM ( = 2*Nsym_primitive + 1) allowed for packing into a 32 bit integer. This is most restrictive at the Translation function stage which needs to store coefficients for all reflection pairs; H-Hj, K-Kj L-Lj where the Hj, Kj, and Lj are symmetry equivalents of H,K,& L. thus needs maximum values for the coefficients which are double the actual ones for the data.
  • See also Memory allocation below concerning possible problems with memory.

KEYWORDED INPUT

The various data control lines are identified by keywords. Only the first 4 characters of a keyword are significant. Records may be continued across line breaks using & or - as the last character on the line to be continued. The available keywords are listed below grouped according to their function:

General Keywords used at any stage:

VERBOSE
produces lots of output.
TITLE
to help you know what you did.

Function keywords:

These call the appropriate procedures.

SORTFUN
calls SORTING procedure to sort and pack reflexions.
TABFUN
calls TABLING procedure to prepare structure factors from the model.
ROTFUN
calls ROTING procedure for the rotation function. (Must be followed by GENE and/or CLMN and/or ROTA).
TRAFUN
calls TRAING procedure for the translation function.
FITFUN
calls FITING procedure for rigid body fitting.
SHIFT
calls REORIENTATE procedure to apply shifts to the model final solution..

Other primary keywords:

May be used for the given functions. 

Keyword         Used in
-------         -------
LABIN           SORTFUN

CRYSTAL         TABFUN, TRAFUN, FITFUN
MODEL           TABFUN
SAMPLE          TABFUN

GENERATE        ROTFUN
CLMN            ROTFUN
ROTATE          ROTFUN

SHIFT           ROTFUN, FITFUN, REORIENTATE
SOLUTION        TRAFUN, FITFUN

SYMMETRY        TRAFUN, FITFUN

REFSOLUTION     FITFUN
END

Subsidiary keywords:

These modify the following primary keywords. Most use sensible defaults.

Keyword       Subsidiary Keywords
-------       -------------------
SORTFUN       RESOLUTION, MODEL
LABIN         FP=??  SIGFP=?? PHI=?? FC=?? PHIC=??

TABFUN        NOROTATE, NOTRANSLATE, NOTAB, HKLOUT, SFOUT
MODEL         BREPLACE, BADD
CRYSTAL       ORTH
SAMPLE        RESOLUTION, SCALE, SHANNON

ROTFUN
GENERATE      RESOLUTION, CELL_MODEL
CLMN          CRYSTAL, MODEL, ORTH, FLIM, SHARPEN, RESOLUTION, SPHERE
ROTATE        CROSS or SELF, MODEL, BESLIM, STEP, PKLIM, NPIC, BMAX, LOCK
SHIFT         COM, EULER

TRAFUN        CB, HL, PT or PTF, CC, NMOL, RESOLUTION, PKLIM, NPIC
CRYSTAL
SYMMETRY
SOLUTION      FIX

FITFUN        NMOL, RESOLUTION, ITER, CONV
REFSOLUTION   AL BE GA X Y Z BF
SHIFT         COM, EULER
CRYSTAL
SYMMETRY
SOLUTION

REORIENTATE
SHIFT         COM, EULER
SOLUTION

SORTFUN keywords

SORTFUN [ RESOLUTION <rmin> <rmax> ] [ MODEL ]

This signals the beginning of Step_1 SORTFUN.

RESOLUTION
<rmin> and <rmax> define the resolution range for all statistics. Can be put in as 4sin(theta)**2/lambda**2 limits, or as Angstrom limits in any order (defaults to MTZ resolution). Data output to HKLPCK0 are restricted to the outer resolution cutoff.
MODEL
This signals that the structure factors input from HKLIN are to be used to make a TABLE. This requires that
they have been calculated from a model placed in a large unit cell and therefore are on a very fine grid.
(See part of example [1]b).

LABIN <column_assignment> ...

[Compulsory.] A line giving the names of the input data items to be selected followed by <program_label>=<file_label> assignments. Acceptable labels are:
FP SIGFP PHI FC PHIC.
FC PHIC must be assigned for structure factors input.
FP must be assigned for creating the list of observations.
If PHI is assigned, the phases are stored and can be used for phased translation searches. 

    LABIN FP=F [ SIGFP=SIGF or PHI=PHIexptl ]
         LABIN FC=FC_domainA PHIC=PHIC_domainA

TABFUN keywords

TABFUN [ NOROTATE ] [ NOTRANSLATE ] [NOTAB] [ HKLOUT ] [ SFOUT ]

This signals the beginning of Step_2 TABFUN

NOROTATE
Do not rotate the model before initialising calculation.
NOTRANSLATE
Do not translate the model before initialising calculation.
Use this extremely rarely. Amore assumes your molecule lies roughly at the origin of the test cell. If you have already run TABFUN, and you wanted to carve pieces out of XYZOUT to do rigid body fitting on segments, it is useful to make a TABLE for each fragment with the TABFUN NOROTATE NOSHIFT option. Similarly if you want to fit another possible model over the first XYZOUT. NEVER use this in an initial pass.
NOTAB
Does not produce a table - just orientate the molecule if appropriate and move the molecule's centre of mass to the origin. This coordinate file can then be used to calculate structure factors and generate Es which can be read in to produce a TABLE file.
HKLOUT
The contents of the TABLE can also be output as an ASCII list of H K L FC PHIC. This may be useful for checking.
SFOUT
An alias for HKLOUT.

MODEL <i> BREPLACE <brep> BADD <badd>

<i> is the model number and is followed by all information needed to work with the model. At least one model must be specified to get any output.

BREPLACE <brep>
replace all B factors in file with <brep>.
Default: Use input B factors
BADD <badd>
add <badd> to all input B factors. If <badd> is negative the `structure factors' are sharpened.
Default: BADD = 0.00

PLEASE NOTE that if all the B-factors are zero in your model, then <badd> MUST be set to a sensible positive value.

The coordinates written to XYZOUT will have the same B-factors as the input coordinates, but the TABLE will be generated using the modified B-factors. Example:
MODEL 1 BREPLACE 0 BADD -10

Other primary keywords (optional):

CRYSTAL <a> <b> <c> <alpha> <beta> <gamma>

Optional. Cell dimensions for observed data used to generate PDB style header for XYZOUT. The default is to use the TABFUN cell to generate the CRYST1 and SCALEi records.

ORTH <i>
orthogonalisation code. See below for conventions. (Default <i>=1.)

Example:
CRYSTAL 112.32 112.32 85.14 90 90 120 ORTH 1

SAMPLE <i> [ RESOLUTION <dmin> ] [ SCALE <scale> ] [ SHANN <sharat> ]

<i> is the model number and is followed by the sampling control parameters.

RESOLUTION <dmin>
<dmin> (in Angstroms) is the resolution limit of generated structure factors. There is no point in setting this higher than the maximum resolution given in SORTFUN.
SCALE <scale>
Optional: default = 4. A model `cell' created equal to (minimal box)*<scale>. This controls how finely the model structure factors are sampled in reciprocal space.
SHANN <sharat>
<sharat> is the Shannon rate for sampling the coordinate map. The default is 2.5. If the B factors have been sharpened it is wise to use a finer grid, i.e. increase <sharat> to 3.5 or 4.

Example:
SAMPLE 1 RESO 3 SHANN 2.5 SCALE 4.0

ROTFUN Keywords {Step_3}

ROTFUN

This signals the beginning of Step_3 ROTFUN with subsequent keywords as follows.

Generate {Step_3a}

GENERATE <i> [ RESOLUTION <rmin> <rmax> ] [ CELL_MODEL <a> <b> <c> ]

<i> is the model number.
This routine calculates the model `structure factors' in a suitable P1 cell, and writes them in the same format as the SORTFUN output for the crystal amplitudes. The file is assigned to HKLPCK1.

RESOLUTION <rmin> <rmax>
Resolution range for data output. Can be put in as 4sin(theta)**2/lambda**2 or as Angstrom limits in either order. Choose the maximum resolution you may wish to use; this step need only be run once for each model and a subset extracted with the resolution limits given in CLMN.
CELL_MODEL <a> <b> <c>
for model structure factor generation. (The angles are always 90 degrees.)
Opinions differ as to the values to use. Eleanor Dodson says: "This model cell needs to be chosen carefully. Ideally you need to use dimensions of Twice maximal distance from Centre of Mass + SPHERE_<Irmax> + a small safety term." She says always use a cubic cell because elongated cells can cause trouble.
Navaza suggests using
{smallest box containing model} + {integration radius (<Irmax>)} + resolution
(not necessarily cubic) and others consider the cell dimensions less critical providing they are chosen large enough to avoid self-vectors.
The maximal distance and minimal box are output by the TABFUN step.

Example:
GENERATE RESO 20 3.2 CELL_MODEL 89 89 89

Calculate spherical harmonics {Step_3b}

CLMN [ CRYSTAL | MODEL <i> ] ORTH <i> FLIM <fmin> <fmax> SHARP <badd> RESO <rmin> <rmax> SPHERE <rmax>

Calculates spherical harmonics for crystal and models.

CRYSTAL
The input is HKLPCK0 for CRYSTAL;
ORTH <i>
orthogonalisation code. (See below for code.) Only needed for CRYSTAL. Except for monoclinic spacegroups with B unique, when ORTH = 3 may be useful, all orthogonalisation codes should be set to 1. Even for the monoclinic case it is usually easier to leave the code as 1. (Default ORTH=1.)
MODEL <i>
HKLPCK1 for MODEL 1.
FLIM <fmin> <fmax>
Minimum and maximum values of F used. (Rarely used option.)
SHARP <badd>
sharpening B value for structure factors This can be used to modify the input F by exp**{-<badd>*sin**2(theta)/lambda**2} before squaring, ie a negative <badd> will sharpen the data
RESOLUTION <rmin> <rmax>
Can be put in as 4sin(theta)**2/lambda**2 or as Angstrom limits in either order. These limits will truncate the H K L listed in HKLPCK. It is important that the SAME resolution limits are used for both the MODEL and the CRYSTAL.
SPHERE <Irmax>
<Irmax> is the radius of the integration sphere in Angstroms. Tips:
  1. This should not be greater than your model's Maximal distance from Centre of Mass output by TABFUN. David Blow points out that for a spherical molecule 75-80% of the molecular diameter includes about 80% of the integrated Patterson density. Ian Tickle suggests using 75% of the minimum diameter in general.
  2. The volume of the sphere should probably not exceed the volume of the asymmetric unit.
  3. If the radius is greater than half the minimum cell edge you will be including some Patterson vectors twice. Opinions differ on how important this is, but the program warns about this case.
Other factors like the shape of the model may influence you; remember this is the RADII within which the interesting self vectors should lie.

Examples: 

CLMN CRYSTAL RESO  20.0  4.0  SPHERE   30  -
                  ORTH  1   SHARP -10.0 FLIM 0.E0 1.E8  
CLMN MODEL 1      RESO  20.0  4.0  SPHERE  30

Rotation {Step_3c}

ROTATE CROSS | SELF MODEL <i> BESLIM <lmin> <lsup> STEP <stepsize> PKLIM <rp> NPIC <np> BMAX <bmax> LOCK <nrot>

This routine calculates the rotation function.

CROSS or SELF
flags whether calculation is to be a SELF rotation, which will only need CLMN0 as input, or a CROSS rotation function which will need CLMN0 and some CLMN<i>. The correlation between self- and cross-rotation functions can be analysed with the program RFCORR.
MODEL <i>
HKLPCK<i> for MODEL <i>.
BESLIM <lmin> <lsup>
Expansion using spherical harmonic functions between <lmin> and <lsup> is done. Low order terms (i.e. for l = 2 or 4) tend to be governed by the crystal symmetry; excluding them may reduce the final peak heights, but make the rotation parameters more precise and make multiple solutions have more equal heights. The upper cut off is governed by the ratio of the integration radius to the resolution. Defaults are 6, 200. The lower cut off has a similar effect to the inner cutoff radius for the Patterson vectors.
STEP <stepsize>
Angular step size for Alpha, Beta and Gamma in degrees (default 2.5). Defaults to sensible value for resolution requested. Should be checked from: STEP ~ 360 / ( 2*<lsup> +1 )
PKLIM <rp>
Output all peaks above <rp> * {maximum peak height}. Default: 0.5 for Cross rotations, 0.2 for self rotations. Maximum self rotation peak will always be the origin peak. The peak search algorithm is not very satisfactory for Beta limits, beta = 0 and beta = <bmax>. Default = 0.5.
NPIC <np>
number of peaks to output. (Limited to 99.)
BMAX <bmax>
Optional: Maximum BETA angle to consider (default 180, or 90 if you have a 2 fold axis perpendicular to the first rotation axis (e.g. in pointgroups Pmmm, P622, P422 etc.).
LOCK <NROT> followed by NROT sets of Eulerian equivalent angles which describe the self rotations.
These control the locked rotation function (see reference [6]).
The Euler angles MUST refer to the SAME orthogonalisation convention as you are using for the CROSS rotation. See example [3]b.

If there are several molecules in the crystal assymmetric unit, AND you know the rotations which relate them to each other, ie you have a solutions to the SELF ROTATION, then the solutions to the cross rotation can be searched to find sets which are related by the expected NCS operators. If you do not have a closed group things are messy. The self rotation always finds pairs of solutions, ie that which rotates Mol1 to Mol2, and that which rotates Mol2 to Mol1. These are the inverse of each other; in Polar coordinates, they have the form (Omega,Phi,Kappa) and (Omega,Phi,-Kappa), and the Eulerian equivalent is (Alpha, Beta, Gamma) and (-Gamma,-Beta, -Alpha).

It is not altogether easy to decide what to do, and you need to have some idea of how many molecules you expect to find in the asymmetric unit, and how they may be arranged. This can be complicated to sort out; if there is a hexamer in the crystal, you would expect to find 3 two-fold axes, all perpendicular to a three fold axis. (If two axes are perpendicular, the product of their direct cosines, 

 DC1(axis1)*DC1(axis2) + DC2(axis1)*DC2(axis2) + DC3(axis1)*DC3(axis2) = 0.0

For TRAP, where the 11-fold rotation axis is perpendicular to a crystallographic 2 fold axis, the self rotation showed both a single peak at (Omega, Phi, 360/11) and 11 2-fold axes. This did NOT mean that TRAP contained 11 dimers, although the self rotation results were consistent with such a conclusion. AMORE does not at present generate all symmetry equivalents of SELF rotation solutions so it is sensible to use MAPROT to give a complete list.

If you believe you have a proper rotation with a clear solution with Kappa equal 360/n, Kappa =180 ( 2-fold), or 120 (3-fold) or 72 (5-fold) and the NCS operators form closed group. then you would specify NROT = n-1, followed by n-1 sets of polar angles to define the rotations: (Omega,Phi,360/n) and (Omega,Phi,2*360/n) etc . In this case, every self rotation solution and its inverse belong to the set.

If say, you expect 222 NCS symmetry with 3 intersecting 2-fold axes, you would set NROT=3 and specify the three sets of two fold axes: (Omega1,Phi1,180), (Omega2,Phi2,180) and (Omega3,Phi3,180).

Example
ROTA CROSS MODEL 1 [ BESLIMI 6 120 STEP 2.5 PKLIM 0.5 NPIC 100]

Reorientation {Step_3d}

SHIFT <Model_number> COM <Xcom> <Ycom> <Zcom> EULER <alpha> <beta> <gamma>

Reorientate stage. Moves Eulerian angle solutions determined for shifted model stored in XYZOUT<Model_number> to give solutions to be applied to original model. Needed if you want your solutions converted back to ones to apply to original coordinates.

COM <Xcom> <Ycom> <Zcom>
coordinates of the molecule's centre of mass output by TABFUN.
EULER <alpha> <beta> <gamma>
rotation angles applied to the original model output by TABFUN.

Example
SHIFT 1 COM 17.3 -10.5 28.7 EULER 301.2 35.7 185.2

TRAFUN keywords {Step_4}

TRAFUN [ CB | CO | PT | PTF | HL | CC ] NMOL <nmol> [ RESOLUTION <rmin> <rmax> ] [ PKLIM <rp> ] [ NPIC <np> ]

There are various translation function targets. Each takes each orientation solution in turn and searches for the NPIC "best" translational Xi Yi Zi for this orientation. Good solutions should give high correlation coefficients between FP and FC, and low Rfactors. Only one target can be specified for each run.

  1. CB | CO - the method of Crowther and Blow (default).
    CB(T) = <DeltaI(obs) * I(calc)(T)>
    The convolution (designated by "*") of the observed Patterson (after subtraction of the contribution of the self vectors) with the calculated one for each value of the translation vector T.
  2. PT | PTF - Phased translation function.
    This can either use externally generated phases for the model (option PTF; input at SORTFUN) or for many body problems phases derived from the FIXed molecules (option PT).
    It looks for the best overlap of the 2 maps: (Fp:PHI model) and (Fc:PHI model). See reference [4].
  3. HL - Harada-Lifchitz.
    HL(T) = <DeltaI(obs) * I(calc)(T)> / < I(calc)(T)>
    Here the convolution has been "normalised".
  4. CC - correlation coefficient.
    CC(T) = <DeltaI(obs) * I(calc)(T)> / sqrt( < DeltaI(obs)**2 * I(calc)(T)**2>
    This function is powerful but much slower.

Each function tests each orientation solution in turn and searches for the best translational Xi Yi Zi for this orientation. Good solutions should give high correlation coefficients between FP and FC, and low Rfactors. For the first molecule all <Xi> <Yi> <Zi> belonging to the Cheshire cell are searched (see reference [7]). The Cheshire cell is the minimum volume which will allow a unique solution. For the first molecule it will be the cell which covers a volume from one possible origin to the next - you can usually see it by inspection of International Tables, e.g.: For P212121, the Cheshire cell is 0-0.5,0-0.5,0-0.5. For P21 the Cheshire cell is 0-0.5,any y,0-0.5. If you are searching for the NMOLth molecule of a set, the Cheshire cell will now be the whole primitive volume. You have assigned the origin by choosing the position of the first molecule, and the other molecules will have to be positioned relative to that choice.

A map of the Cheshire cell for each search is written to the file assigned to MAPOUT. N.B. the same file is used for all solutions - only the final one will be saved. If you wish to plot your best solution you will have to recalculate it.

Translation functions use a great deal of memory. The whole FFT transform is held in memory at once, and the calculation is done over a set of reciprocal lattice coefficients which can be twice the size of Hmax, Kmax, Lmax.

NMOL <nmol>
Number of molecules to search for (maximum 9). The program assumes you have solutions for <nmol>-1 molecules and searches for the best fit for the <nmol>-th one. The <nmol>-1 solutions must be FIXed; see examples [6], [7], [8]. Default = 1.
RESOLUTION <rmin> <rmax>
Can be put in as 4sin(theta)**2/lambda**2 or as Angstrom limits in any order.
PKLIM <rp>
Output all peaks above <rp*>{maximum peak height}. Default 0.5.
NPIC <np>
Number of peaks to output from the translation function map for each orientation. Default 10. Be aware that the highest peaks in the translation function map do not necessarily correspond to the highest correlation coefficients. All targets are prone to generate "noise" peaks, and good solutions usually satisfy all 3 criteria: High T1 peak, high correlation coefficient, low Rfactor.

Example
TRAFUN CO NMOL 1 RESO 8 4 PKLIM 0.5 NPIC 10

Other optional keywords

SYMMETRY <spg>

(Optional.) Spacegroup name or spacegroup number It will default to that of the CRYSTAL data, picked up at the SORTFUN step. You may need to change it to test other possibilities; e.g. enantiomorphic spacegroups - P65 instead of P61. If you are not sure of your spacegroup, the translation function is a good way to distinguish the true spacegroup; e.g. you may need to test all possible orthorhombic possibilities; P222; P2 2 21; P2 21 2; P2 21 21; P21 2 2; P 21 2 21; P21 21 2; P 21 21 21; See example [4], [5].

CRYSTAL FLIM <fmin> <fmax> ORTH <i> SHARP <badd> RESOLUTION <rmin> <rmax>

(Optional.) Information used to modify the CRYSTAL amplitudes. See descriptions above for CLMN. Example:
CRYSTAL ORTH 1 FLIMI 0.E0 1.E8 SHARP 0.0

Other compulsory keywords

SOLUTION [FIX] <i> <alphai> <betai> <gammai> [ <Xi> <Yi> <Zi> ]

When searching for a single molecule, a list of possible orientations from the rotation function (labelled SOLUTIONRC in ROTFUN output) is required.

Molecules are found sequentially. When searching for the nth molecule of a set, there must be sets of (n-1) previously determined solutions to the translation function. These are labelled with the key word FIX. For example to find the 2nd molecule fix one solution: 

SOLUTIONTF1 FIX 1 <alpha1> <beta1> <gamma1>  <X1> <Y1> <Z1>
followed by the set of possible rotation function solutions. Each rotation orientation is tested in turn with the previous input FIXed solution. If you want to test several translation solutions, you can repeat the FIX information, and again follow it with the set of possible rotation function solutions.

To find the 3rd molecule fix a pair of solutions: 

SOLUTIONTF1 FIX 1 <alpha1> <beta1> <gamma1>  <X1> <Y1> <Z1>
SOLUTIONTF2 FIX 1 <alpha2> <beta2> <gamma2>  <X2> <Y2> <Z2>
There is a limit of 99 (calculated as NMOL* Number_of_solutionrc) on the number of orientation solutions which can be included in one run. However there is no extra overhead in submitting several runs. This list should come last and is terminated by end-of-file or the keyword END.
The list of solutions can be extracted from ROTFUN (and TRAFUN) output using grep and edited in here.

<i>
<i> is the number for the appropriate TABLE<i>.
<alphai> <betai> <gammai>
Euler angles output by ROTFUN. If there are no clear maxima you should test many solutions. Correct solutions have been found from rotation solutions which were far down the list.
FIX <alphai> <betai> <gammai> <Xi> <Yi> <Zi>
If the molecule generated by this solution is FIXed these 6 parameters define its position in the cell. Structure factors calculated from this molecule will be added to those generated for molecules which are being searched for.

Examples
SOLUTIONTF FIX 1 27.8 100.7 350.1 0.146 0.566 0.00 17.4 52.5
SOLUTIONRC 1 25.211 105.573 339.440

HINTS

To extract the rotation information, `grep' (Unix) or `SEARCH' (VMS) for `SOLUTIONRC' in the ROTFUN output. Edit the resulting list to include only those solutions you want to run the translation search on, and include them in the input data e.g. with `@<file>'.

If you are searching for the <nmol>th molecule of a set, you must FIX <nmol>-1 solutions and search for the <nmol>th one. You will probably have several sets of the fixed solutions to test, plus many possible orientation solutions.

FIXed solutions will be extracted from your previous TRAFUN log. They will be followed by the list of solutions to the Rotation function output by Step_3. Structure factors calculated from the FIXed solutions are added to those generated for search molecules.

To extract the information for FIXed grep for `SOLUTIONTF'.
You will need to sort these to find those with the highest correlation coefficients, and lowest Rfactors. 

 sort -r +8 -9 tra.list > tra_cc.list  # sort on correlation coefficient.
 sort +9 -10 tra.list > tra_rf.list  # sort on Rfactor

(Be careful to keep sets of solutions together!)

See the Unix plumbing in the example scripts, e.g., `auto-amore'.

FITFUN keywords {Step_5}

FITFUN

This signals the beginning of Step_5 FITFUN which performs Rigid-body refinement. It minimises the sum over all hkl of ({Fo*exp(-Bs**2)}**2 - {k*Fc**2})**2 with respect to scale, B-factor and rotation and translation parameters.

Subsidiary words after FITFUN: (many same as TRAFUN)

NMOL <nmol>
Number of molecules to fit. All are fitted together by an iterative procedure.
RESOLUTION <rmin> <rmax>
Can be put in as 4sin(theta)**2/lambda**2 or as Angstrom limits in any order. Often sensible to "fit" the molecules against high resolution data if the sequence homology is close.
ITER <niter>
number of iterations (default 10)
CONV <con>
convergence acceptance (default 0.001)

Example
FITFUN NMOL 3 RESO 20 4.5 ITER 10 CONV 1.E-3

Extra keywords

CRYSTAL FLIM <fmin> <fmax> ORTH <i> SHARP <badd> RESOLUTION <rmin> <rmax>

(Optional.) Information used to modify the CRYSTAL amplitudes. See descriptions above for CLMN.

SYMM <spg>

(Optional.) Spacegroup name or spacegroup number. It will default to that of the CRYSTAL data, picked up at the SORTFUN step. You may need to change it to test other possibilities; e.g. enantiomorphic spacegroups - P65 instead of P61.

REFSOLUTION [ BF ] [ AL ] [ BE ] [ GA ] [ X ] [ Y ] [ Z ]

Refinement to be done for any of temperature factor, alpha, beta, gamma, x, y, z. Remember - in polar spacegroups you cannot refine either y or z parameter for one solution. This defaults to sensible values for different space groups.

Optional: program chooses sensible defaults.

Example
REFSOL AL BE GA X Y Z BF

SOLUTION <i> <alphai> <betai> <gammai> [ <Xi> <Yi> <Zi> ]

<i>
model number for input. Different solutions may require different model numbers. Assign all TABLE<i>.
<alphai> <betai> <gammai>
Euler angles output by ROTFUN. If there is no clear maximum you should test many solutions. Correct solutions have been found from rotation solutions which were far down the list.
<Xi> <Yi> <Zi> [ <CCi> <RFi> ]
These three parameters define the molecules position in the cell. It is often convenient to keep the correlation coefficient and R factor on the solution line. It helps to monitor solutions - subsequent steps should improve these parameters!. The solutions are refined in sets of NMOL. There may be up to 99 solutions given (99/NMOL sets).

Examples 

SOLUTIONTF 1  25.1  105.6  339.5 0.1139  0.5691  0.0000
SOLUTIONTF 1  27.6  100.6 350.3 0.1461 0.5716 0.6476 48 51
SOLUTIONTF 1  27.7  115.9 353.5 0.1439 0.6027 0.3584 49 54

This list is terminated by end-of-file or the keyword END.

This list of Eulerian angles and translations can be extracted from the log file and edited in here. To extract the information from the previous log file, grep for `SOLUTIONTF'. You will need to sort these to find those with the highest correlation coefficients, and lowest Rfactors as described in step_4a, and edit to include only those solutions you want to run the rigid body refinement on to include them in the input data.

SHIFT <Model_number> COM <Xcom> <Ycom> <Zcom> EULER <alpha> <beta> <gamma>

Reorientate stage. Moves Eulerian angle solutions determined for shifted model stored in XYZOUT<i> to give solutions to be applied to original MODEL. Needed if you want your solutions converted back to ones to apply to original coordinates.

COM <Xcom> <Ycom> <Zcom>
coordinates of the molecules centre of mass output by TABFUN
EULER <alpha> <beta> <gamma>
rotation angles applied to the original model output by TABFUN.

Example
SHIFT 1 COM 17.3 -10.5 28.7 EULER 301.2 35.7 185.2

REORIENTATE keywords {Step_6}

SHIFT <Model_number> COM <Xcom> <Ycom> <Zcom> EULER <alpha> <beta> <gamma>

This signals the beginning of Step_6 - reorientate stage. This step can be run as a standalone step or as part of ROTFUN or FITFUN. It moves Eulerian angle solutions determined for shifted model stored in XYZOUT<i> to give solutions to be applied to original MODEL. Needed if you want your solutions converted back to ones to apply to original coordinates.

COM <Xcom> <Ycom> <Zcom>
coordinates of the molecule's centre of mass output by TABFUN
EULER <alpha> <beta> <gamma>
rotation angles applied to the original model output by TABFUN.

Example
SHIFT 1 COM 17.3 -10.5 28.7 EULER 301.2 35.7 185.2

Compulsory following keyword

SOLUTION <i> <alphai> <betai> <gammai> <Xi> <Yi> <Zi>

There may be up to 99 solutions given. This list is terminated by end-of-file or the keyword END.

Examples 

SOLUTIONTF 1  25.1  105.6 339.5 0.1139 0.5691 0.0000
SOLUTIONTF 1  27.6  100.6 350.3 0.1461 0.5716 0.6476 43.5 46.5
SOLUTIONTF 1  27.7  115.9 353.5 0.1439 0.6027 0.3584 41.3 47.3

END

Must be last keyword. Used as termination for list of solutions.

NOTES

Memory allocation

The program currently uses a lot of memory. At several points a whole Fourier transform is held in memory, and it is easy to overflow the limits set. You may not have enough virtual memory available to run large cases. It should be made more memory-efficient in the future. In the meantime it does dynamic memory allocation; the amount allocated at runtime is parameterised by assigning values to logical names since it currently isn't able to compute how much is needed at each stage before doing the allocation. Thus there may be some trial and error involved in setting appropriate values. The defaults are chosen to allow solutions of realistic cases on a `typical' VAX system. Note that the amount of memory you can grab may depend on what else the system is doing as well as possible per-user limits, so it may pay to try later on a multi-user system.

If the allocation for an array isn't large enough, the program stops with a message which should indicate at least which parameter needs to be increased and, in most cases, to what value. If the message doesn't make it clear what needs to be increased, please report the fact. Using the keyword VERBOSE may give more indication. The current values are printed in the output (look for `Memory allocation'). They may be changed by giving the appropriate logical names an integer value (which represents the size of an array) in any of possible ways:

  • On the command line e.g., `TABLING_MR 5400000';
  • From the environment:
    DEFINE TABLING_MR 5400000 ! VMS
    setenv TABLING_MR 5400000 # csh
    TABLING_MR=5400000 # sh
  • By editing $CINCL/default.def e.g. with a line:
    TABLING_MR=5400000

The last option may be most appropriate on a system with lots of memory to provide large defaults and the distributed default.def contains commented-out values for a `big' version used at York and Cambridge.

Rotation matrix definitions

The convention is that the orthogonalised coordinates of "crystal 2" (usually the model) are rotated to overlap the orthogonalised coordinates of crystal 1. 

i.e. [XO1]    = [ROT] [XO2]
     [YO1]            [YO2]
     [ZO1]            [ZO2]
This means that axis permutations introduced by using NCODE = 2, 3 or 4 will result in apparently different solutions, although the effect on the fractional coordinates is the same.

In Polar angles:

If l m n are the direction cosines of the axis about which the rotation k = kappa takes place, and: 

( l )    ( sin omega cos phi )
( m )  = ( sin omega sin phi )
( n )    ( cos omega )
where omega is the angle the rotation axis makes to the ZO direction, and phi is the angle the projection of the rotation axis onto the XO-YO plane makes to the XO axis. 

 [ROT] =
( l**2+(m**2+n**2)cos k     lm(1-cos k)-nsin k        nl(1-cos k)+msin k   )
( lm(1-cos k)+nsin k        m**2+(l**2+n**2)cos k     mn(1-cos k)-lsin k   )
( nl(1-cos k)-msin k        mn(1-cos k)+lsin k        n*2+(l**2+m**2)cos k )
Note that if omega = 0 or 180, then phi is indeterminate and is flagged as 999 in the SOLUTIONs output by AMORE.

In Eulerian angles:

If a (alpha) represents a rotation about the initial ZO axis,
b (beta) represents a rotation about the new position of the YO axis, and
g (gamma) represents a rotation about the final ZO axis: 

 [ROT] =
( cosa cosb cosg - sina sing     -cosa cosb sing - sina cosg     cosa sinb )
( sina cosb cosg + cosa sing     -sina cosb sing + cosa cosg     sina sinb )
( -sinb cosg                     sinb sing                       cosb      )

Orthogonalisation codes

orthogonalisation code NCODE
   = 1, orthogonal x y z along a,c*xa,c* (Brookhaven, default)
   = 2                         b,a*xb,a*
   = 3                         c,b*xc,b*
   = 4                         a+b,c*x(a+b),c*
   = 5                         a*,cxa*,c   (Rollett)

EXAMPLES

  1. #    sorting run:
#   #############
#


    1. # MTZ file contains cell and symmetry.
#
amore hklin spmi_trun.mtz hklpck0 spmipch.hkl sorting_nr 1000000 <<  eof
TITLE   ** spmi  packing h k l F for crystal**
SORTFUN RESOL 100.  3.
LABI FP=F  SIGFP=SIGF
eof

      2. 

    2. # Converting structure factors generated from a blob of electron density
# to a TABLE. The blob must have been placed in a large "P1 unit cell".
#

#!/bin/csh -f
###########################################################
#
# There are lots of alternative ways of getting a masked block of density.
#  You first need a mask.
# This is the simplest technique I have used..
#
# Another way is to edit bones, then use bones_to_pdb to write out a file of
# coordinates, and use ncsmask with that set, and the default atom radius.
#  ( 3A I think..)
#
####################################################################
#  Make a spherical mask centred at the centroid of the chosen block of 
#  density.
#  You need to choose a volume completely contained within the P1 cell; 
#  ie all parts have coordinates between 0 and 1.
# This is important later on for the amore translation.
#   By choosing the right symmetry operator, I have always managed to do 
#  this.. although sometimes the block radius has had to be restricted a bit.
#  This doesnt seem to matter - you will have most of the molecular volume..
###########################################################
#  P65_block_com.pdb 
# REMARK COM of a pva block - 18A radii 
#REMARK X: 22to55/103 Y; 22to62/102 Z; 60 to 89/96 
#CRYSTL  208.400  208.400   96.200  90.00  90.00 120.00    P65
#SCALE1      0.004798  0.002770  0.000000       0.000000
#SCALE2      0.000000  0.005541  0.000000       0.000000
#SCALE3      0.000000  0.000000  0.010395       0.000000
#ATOM xcent Ycent Zcent
#
ncsmask xyzin ./P65_block_com.pdb \
mskout $SCRATCH/P65_block_com.msk <<eof
#  I have taken a 1A grid.
GRID  204  204   96
AXIS   Y    X    Z
RADIUS 18
END
eof
#
###########################################################
# extend the  DM map to the same limits as the msk;
#   you will have to look at the log of Step 1.
#  ( You can get the mask grid by typing 
#     prmap mapin $SCRATCH/P65_block_com.msk )
###########################################################
mapmask mapin /y/work2/suresh//nat3_au5_hg2_dm.map \
mapout $SCRATCH//nat3_au5_hg2_dm.ext << eof
XYZLIM  57  93    62 101    56 91
END
eof
#
###########################################################
#    Generate a pseudo map in a big cell to act as a "model" for maprot 
#    you will want to generate a list of structure factors 
#    on a fine grid for Amore, and this requires a big cell.
#    There must be other ways of doing it but this works..
#
# You will have to choose this cell sensibly, look at other amore
# TABFUN outputs for guidance
# Must be at least double the density block size
#
#  bigdummycell.pdb  - a dummy cell with only one atom
#  CRYSTL  120.000  120.000  120.000  90.00  90.00  90.00    1
#  REMARK CRYSTAL 259.992  250.904  125.504  90.00  90.00  90.00
#  ATOM      1  CB  ALA    13  1    1.974   3.548   9.307  1.00 61.57   6
#
sfall xyzin ./bigdummycell.pdb \
mapout $SCRATCH/bigdummymap.map <<eof
MODE ATMMAP
#SCALE 0.0
SYMM P1
GRID 300 300 300
END
eof
#
###########################################################
#   Now the tricky bit - put the "good" density in the big P1 cell:
#   This takes a lot of core and crashes my little Indy!
#
maprot \
mapin  $SCRATCH/bigdummymap.map \
wrkin $SCRATCH//nat3_au5_hg2_dm.ext \
mskin $SCRATCH/P65_block_com.msk \
mapout $SCRATCH/nat3_au5_hg2_dm_cent_bigdummycell.map \
<<eof
# "MODE TO" moves the WRKIN map ( after masking with MSKIN) to the MAPIN grid.
MODE TO
#  No averaging; this is the identity..
SYMM P1
AVERAGE 1
ROTATE EULER 0 0 0
TRANS 0 0 0
END
eof
#
###########################################################
#
#  Generate structure factors from this density ready for Amore
# Then delete the *bigdummy*maps - they are HUGE..
sfall \
mapin  $SCRATCH/nat3_au5_hg2_dm_cent_bigdummycell.map \
hklout $SCRATCH/nat3_au5_hg2_dm_cent_bigdummycell.mtz \
<<eof
MODE SFCALC MAPIN
SYMM P1
RESO 37 2.5
LABO FC=FC1 PHIC=PHIC1
END
eof
#
#  Now run new Amore to read these SFS in and generatethe TABLE
#####################################################3
#    sorting run:
#####################################################3
#  mtz file contains cell and symmetry.
#
amore \
hklin $SCRATCH/nat3_au5_hg2_dm_cent_bigdummycell.mtz \
table1 $SCRATCH/nat3_au5_hg2_dm_cent_bigdummycell.tab  \
<<'END'
VERBOSE
TITLE   ** packing h k l F for crystal**
SORTFUN MODEL 100 3
LABI FC=FC1  PHIC=PHIC1
'END'
#
# And on the ROTFUN - this step has replaced the TABFUN step


    2. 

  2. #    tabling run:
#   #############
#
#  rotate and shift coordinates and produce table:
#    xyzout is the rotated and shifted coordinates.
#
amore  xyzin1  search.pdb xyzout1 searchrot.pdb \
       TABLE1 search.tab tabling_mi 10000000 \
       tabling_mr 10000000 tabling_mc 1000000 << eof
TITLE :  Produce table for MODEL FRAGMENT
VERBOSE
TABFUN
CRYSTAL  112.32 112.32 85.14 90 90 120 ORTH 1
MODEL 1 BREPLACE 0 BADD 0
SAMPLE 1 RESO 3 SHANN 2.5 SCALE       4.0
eof
    3. 


  3. #    roting run:
#   ############
#

      

    1. #  straightforward rotation function.
#
amore  TABLE1 search.tab \
       HKLPCK1 search.hkl \
       hklpck0 spmipch.hkl \
       clmn1 search.clmn \
       clmn0 spmipch.clmn  \
       roting_mi 1000000 \
       roting_mr 1000000 \
       roting_mc 10000000 \
       roting_md 100000 \
       MAPOUT amore_cross.map <<  eof
ROTFUN
VERB
TITLE : Generate HKLPCK1 from MODEL FRAGMENT   1
GENE 1   RESO 100.0 3.0  CELL_MODEL 80 75 65
CLMN CRYSTAL ORTH  1 RESO  20.0  4.0  SPHERE  0.0 30
CLMN MODEL 1     RESO  20.0  4.0 SPHERE  0.0 30
ROTA  CROSS  MODEL 1  PKLIM 0.5  NPIC 100
eof
      2. 


    2. #  reorientate with PDBSET.
Assume the following three solutions from AMoRe:
# SOLUTIONF     1   56.35   74.98  145.14  0.3883 -0.0061  0.2757 55.7 45.2 57.1  28
# SOLUTIONF     1  295.44   70.84  148.61  0.8273  0.9301  0.2737 55.7 45.2 57.1  29
# SOLUTIONF     1  164.23   69.22  147.81  0.0896  0.8444  0.2876 55.7 45.2 57.1  30
Then:
pdbset \
xyzin /y/ccp4/work/model-rot.pdb \
xyzout /y/ccp4/work/model-rot-sol1.pdb \
<<eof
CELL 78.700   40.400   56.000  90.00 117.10  90.00 
SYMM C2
rotat euler   56.35   74.98  145.14  
shift frac 0.3883 -0.0061  0.2757 55.7 45.2 57.1  28
chain A
end
eof
#
pdbset \
xyzin /y/ccp4/work/model-rot.pdb \
xyzout /y/ccp4/work/model-rot-sol2.pdb \
<<eof
# Use -0.5,-0.5,0 = other C2 solution
CELL 78.700   40.400   56.000  90.00 117.10  90.00 
SYMM C2
rotat euler  295.44   70.84  148.61
shift frac 0.3273  0.4301  0.2737 55.7 45.2 57.1  29
chain B
end
eof
#
pdbset \
xyzin /y/ccp4/work/model-rot.pdb \
xyzout /y/ccp4/work/model-rot-sol3.pdb \
<<eof
#  Subtract 1 from y
CELL 78.700   40.400   56.000  90.00 117.10  90.00 
SYMM C2
rotat euler  164.23   69.22  147.81
shift frac 0.0896 -0.1556  0.2876 55.7 45.2 57.1  30
chain C
end
eof
#

#
    4. 

  4. #    traing run:   NMOL = 1 - P61
#   #############################
#
amore  TABLE1 search.tab \
       HKLPCK0 spmipch.hkl \
       traing_nr 100000 \
       traing_meq 100 \
       traing_mrt 10000000 \
       traing_mt 10000000 \
       traing_mr 10000000 \
       MAPOUT amore_transjunk1.map <<  eof
TRAFUN CB   NMOL 1 RESO 8 4  PKLIM 0.5  NPIC 10
SYMM P61
VERB
TITLE : Translation function P61 - one molecule
SOLUTIONRC 1    25.211   105.573   339.440
SOLUTIONRC 1    27.757   100.743   350.082
SOLUTIONRC 1    27.939   115.792   353.601
SOLUTIONRC 1    27.596    60.308    43.149
SOLUTIONRC 1    38.604    77.537   160.999
SOLUTIONRC 1    16.079   130.379   261.311
SOLUTIONRC 1     7.264    66.987    88.523
SOLUTIONRC 1     4.345    82.989    95.253
SOLUTIONRC 1    26.903    76.829    37.613
SOLUTIONRC 1     1.477    33.145    73.636
SOLUTIONRC 1    42.057   104.775   163.088
SOLUTIONRC 1     0.492    90.289   275.552
SOLUTIONRC 1    53.344   135.528   269.211
SOLUTIONRC 1    34.118    74.264   244.711
SOLUTIONRC 1    42.237   147.472   263.153
SOLUTIONRC 1    33.968     5.665   291.432
eof
    5. 


  5. #    traing run:   NMOL = 1 - P65
#   #############################
#
amore  TABLE1 search.tab \
       HKLPCK0 spmipch.hkl \
       traing_nr 100000 \
       traing_meq 100 \
       traing_mrt 10000000 \
       traing_mt 10000000 \
       traing_mr 10000000 \
       MAPOUT amore_transjunk5.map <<  eof
TRAFUN CB   NMOL 1 RESO 8 4  PKLIM 0.5  NPIC 10
SYMM P65
VERB
TITLE : Translation function P65 - one molecule
SOLUTIONRC 1    25.211   105.573   339.440
SOLUTIONRC 1    27.757   100.743   350.082
SOLUTIONRC 1    27.939   115.792   353.601
SOLUTIONRC 1    27.596    60.308    43.149
SOLUTIONRC 1    38.604    77.537   160.999
SOLUTIONRC 1    16.079   130.379   261.311
SOLUTIONRC 1     7.264    66.987    88.523
SOLUTIONRC 1     4.345    82.989    95.253
SOLUTIONRC 1    26.903    76.829    37.613
SOLUTIONRC 1     1.477    33.145    73.636
SOLUTIONRC 1    42.057   104.775   163.088
SOLUTIONRC 1     0.492    90.289   275.552
SOLUTIONRC 1    53.344   135.528   269.211
SOLUTIONRC 1    34.118    74.264   244.711
SOLUTIONRC 1    42.237   147.472   263.153
SOLUTIONRC 1    33.968     5.665   291.432
eof
    6. 


  6. #    traing run:   NMOL = 2 - P61
#   #############################
#
amore  TABLE1 search.tab  \
       traing_nr 100000 \
       traing_meq 100 \
       traing_mrt 10000000 \
       traing_mt 10000000 \
       traing_mr 10000000 \
       HKLPCK0 spmipch.hkl <<  eof
TRAFUN CB   NMOL 2 RESO 8 4  PKLIM 0.5  NPIC 10
SYMM P61
VERB
TITLE : Translation function P61  - 2 mols together.
SOLUTIONTF FIX 1    27.76   100.74   350.08 -
   0.14596  0.56602  0.00000 17.4 52.5
SOLUTIONRC 1    27.94   115.80   353.60
SOLUTIONRC 1    25.21   105.57   339.45
SOLUTIONRC 1    27.94   115.80   353.60
SOLUTIONRC 1    27.76   100.74   350.08
eof
    7. 


  7. #    traing run:   NMOL = 2 - P65
#   #############################
#
amore  TABLE1 search.tab  \
       traing_nr 100000 \
       traing_meq 100 \
       traing_mrt 10000000 \
       traing_mt 10000000 \
       traing_mr 10000000 \
       HKLPCK0 spmipch.hkl <<  eof
TRAFUN CB   NMOL 2 RESO 8 4  PKLIM 0.5  NPIC 10
SYMM P65
VERB
TITLE : Translation function P65  - 2 mols together.
SOLUTIONTF FIX 1    27.76   100.74   350.08 -
 0.14596  0.56602  0.00000 17.4 52.5
SOLUTIONRC 1    27.94   115.80   353.60
SOLUTIONRC 1    25.21   105.57   339.45
SOLUTIONRC 1    27.94   115.80   353.60
SOLUTIONRC 1    27.76   100.74   350.08
eof
    8. 


  8. #  traing run:   NMOL = 3 - P65
#   ###########################
#
#  (no point in testing P61 now - obv P65 better)
#
amore  TABLE1 search.tab  \
       HKLPCK0 spmipch.hkl  \
       traing_nr 100000 \
       traing_meq 100 \
       traing_mrt 10000000 \
       traing_mt 10000000 \
       traing_mr 10000000 \
       TRAFUN trafun.9 <<  eof
TRAFUN CB   NMOL 3 RESO 8 4   PKLIM 0.5  NPIC 10
SYMM P65
VERB
TITLE : Translation function P65  - 2 mols together.
SOLUTIONTF FIX 1    25.21   105.57   339.45 -
 0.11330  0.56704  0.00000 38.0 46.7
SOLUTIONTF FIX 1    27.76   100.74   350.08 -
 0.14660  0.57107  0.65289 38.0 46.7
SOLUTIONRC 1    27.94   115.80   353.60
SOLUTIONTF FIX 1    25.21   105.57   339.45 -
 0.11146  0.56738  0.00000 35.8 47.0
SOLUTIONTF FIX 1    27.94   115.80   353.60 -
 0.14490  0.60324  0.35856 35.8 47.0
SOLUTIONRC 1    27.76   100.74   350.08
SOLUTIONTF FIX 1    27.76   100.74   350.08 -
 0.14596  0.56602  0.00000 31.3 48.8
SOLUTIONTF FIX 1    27.94   115.80   353.60 -
 0.14472  0.60356  0.70544 31.3 48.8
SOLUTIONRC 1    25.21   105.57   339.45
eof
    9. 


  9. #    fiting run:
#   ############
#
amore  TABLE1 search.tab  \
       fiting_meq 100 \
       fiting_mt 10000000 \
       fiting_nr 100000 \
       fiting_np 10 \
       HKLPCK0 spmipch.hkl <<eof
FITFUN  NMOL 3  RESO 20 4.5  
TITLE *** spmi   structure ***
VERBOSE
REFSOL   AL     BE   GA     X   Y    Z   BF
SOLUTIONTF 1  25.02  105.58  339.46 0.11386  0.56908  0.00000
SOLUTIONTF 1  27.60  100.60  350.29 0.14607 0.57157 0.64759 43.5 46.5
SOLUTIONTF 1  27.72  115.95  353.54 0.14386 0.60270 0.35841 41.3 47.3
eof

AUTHORS

Jorge Navaza. Adapted for CCP4 by Eleanor Dodson.

REFERENCES

  1. J.Navaza, Acta Cryst. A50, 157-163 (1994)
    (General reference.)
  2. J.Navaza. Acta Cryst. A43, 645-653 (1987)
    (Radial quadrature instead of bessel expansion)
  3. J.Navaza. Acta Cryst. A46, 619-620 (1990)
    (Stable recurrence relationship for rotation matrices.)
  4. G.A.Bentley, Some applications of the phased translation function using calculated phases in Molecular Replacement, Proceedings of the Daresbury Study Weekend, (1992) DL/SCI/R33
  5. E.E.Castellano et al., Fast Rigid-body Refinement for Molecular-replacement Techniques, J. Appl. Cryst. 25, 281-4 (1992).
  6. J.Navaza. Acta Cryst. D49, 588-591 (1993)
  7. Hirschfeld Acta Cryst. A24, 301-311 (1968)

SEE ALSO

almn, ecalc, lsqkab, npo, pdbset, rfcorr