ANISOANL (CCP4: Supported Program)

NAME

anisoanl - analyses of anisotropic displacement parameters

SYNOPSIS

anisoanl xyzin input.pdb tlsin input.tls xyzout output.pdb tlsout output.tls psout out.ps
[Key-worded input file]

DESCRIPTION

The program analyses model anisotropic U values supplied on ANISOU records in the input PDB file (XYZIN). These may originate for example from a full anisotropic refinement by REFMAC for atomic resolution data. Plots of average equivalent isotropic B values, anisotropy, and radial and tangential projections against residue are produced.

Using rigid groups defined in TLSIN, it can fit TLS values to the observed U values. These are output in TLSOUT, and residual U values are output in XYZOUT. The so-called L2 norm is used as the residual for fitting U's calculated from TLS tensors to the observed Us obtained from refinement.

The program can also analyse the input anisotropic U values in terms of Rosenfield's rigid-body postulate. Output plots give an indication of whether groups of atoms (as defined in TLSIN) have U values conforming to rigid-body-like displacements. A postscript plot is also produced which may hint at possible rigid groups.

The plots against residue may be useful for visualising U values obtained from the program TLSANL. However, the rigid group analysis is less useful, since in this case the U values will have been obtained from a rigid group description in the first place.

INPUT AND OUTPUT FILES

XYZIN

Input coordinates with anisotropic U values held in standard ANISOU records. The elements of U are assumed to appear as integers representing 10000*Uij in orthogonal coordinates, and in the order U11, U22, U23, U12, U13, U23.

The program will check for non-positive-definite anisotropic U values, and report any found to the log file. Non-positive-definite means that one of the eigenvalues is less than or equal to zero, which in turn means that one of the radii of the thermal ellipsoid has vanished.

TLSIN

Rigid group definitions for TLS groups. The format of the TLS file is as given in the RESTRAIN documentation, with the following addition. The RANGE record can include the subkeyword FIT or APPLY. Atoms included in the FIT range are used both for fitting the TLS values, and in calculating residual U's for XYZOUT. Atoms in APPLY are used only for the latter. Thus, one may want to use only main chain atoms for fitting TLS values, but then apply these TLS values to all atoms in the molecule. Only TLS and RANGE records are required, to define the chosen TLS groups.

XYZOUT

Output coordinates and residual anisotropic U values.

TLSOUT

Fitted TLS tensors for groups defined in TLSIN. This file can be input to TLSANL for further analysis. Note that TLSOUT contains a REFMAC record to flag that the file contains tensor elements in the order used in REFMAC, rather than the order used by RESTRAIN.

PSOUT

Postscript plot of "delta" values between all selected pairs of atoms. Light shading implies low "delta" value, consistent with the atoms belonging to the same quasi-rigid group. Dark shading means the atoms are unlikely to belong to the same quasi-rigid group. See RIGIDBODY keyword.

KEYWORDED INPUT

RIGIDBODY [OFF]

Assess Rosenfield's rigid-body postulate (default). If OFF specified then don't. The rigid-body postulate states that for two atoms belonging to the same rigid-body (not necessarily bonded), the projections of the anisotropic displacement parameters along the interatomic vector should be equal. In practice, we expect the difference between the projections to be smaller for atoms belonging to the same quasi-rigid body, and larger for atoms belonging to different quasi-rigid bodies.

For all pairs of atoms, the absolute difference in the projection as a fraction of the mean projection (the "delta" value) is calculated, and all such differences are binned (see the keyword DUBINS). These "delta" values are displayed graphically in a postscript plot (file PSOUT, default anisoanl.ps). Light shading implies low "delta" value, consistent with the atoms belonging to the same quasi-rigid group. Dark shading means the atoms are unlikely to belong to the same quasi-rigid group. Atom selection can be done with the file TLSIN - only atoms specified in this file are used in the calculation. For example, clearer results may be obtained if only CA atoms are used. See Tom Schneider's article for an example of this kind of plot.

The distribution of "delta" values is included in the log output (see keywords DUBINS and DURANGE). Possible quasi-rigid bodies should be defined using the TLSIN file (see example below). The distribution is plotted for all pairs of atoms within each quasi-rigid body, and a final plot gives the distribution for pairs of atoms from different groups. If the choice of rigid bodies is good, the differences should be significantly smaller within groups than between them. A subset of atoms can be chosen using the atom selection field in TLSIN (e.g. "CA" may be useful for large rigid groups).

DUBINS <dist_bins> <ps_bins>

Specify the number of bins for the RIGIDBODY plots. <dist_bins> refers to the distribution plot, and <ps_bins> to the postscript plot. Defaults are 30 and 10 (maximum is 100).

DURANGE <range>

Specify the range of values covered by the RIGIDBODY plot. Default is 0.3. The "delta" values are defined as the absolute difference in the projection as a fraction of the mean projection, so the maximum range is 2.0

PLOT [OFF]

Produce plots of equivalent isotropic U values, anisotropy, and projections of U (default). If OFF specified then don't. See below for a description of the plots produced.

MAINCHAIN

Use main chain atoms only in producing plots. Default is to use all atoms.

VERBOSE

Produces extra output.

END

Terminate input.

PROGRAM OUTPUT

FITTLS option

For each cycle of fitting, the residual, the R value and the Goodness of Fit are printed. The R value is sqrt(sum deltaU**2 / sum Uobs**2), and the Goodness of Fit is 1000 * sqrt(sum deltaU**2 / (num observables - num parameters)) (the factor of 1000 is one over a typical sigmaU). These values are given for (i) atoms used in fitting (FIT), (ii) atoms included in TLS group but not used for fitting (APPLY).

PLOT option

The PLOT option displays graphs of the following quantities:

Uiso: The equivalent isotropic U factor calculated as 1/3*trace(U).
R2FROMORIG: The square of the distance from the local origin. If the FITTLS option is being used, then the local origin is taken to be the origin of the TLS group to which the atom belongs. Otherwise, the local origin is taken to be the centroid of the chain to which the atom belongs (i.e. the mean atomic coordinates of that chain). The values of R2FROMORIG are divided by a scaling factor (currently 3000) for convenience of plotting.
Anisotropy: This is defined as the ratio of the smallest to the largest eigenvalue of U.
PROLMEAN: This factor is defined as the ratio of the middle to the largest eigenvalue of U. If the thermal ellipsoid corresponding to U is oblate (disc-like), then this factor will be close to 1. If however it is prolate (cigar-like), then this factor will be close to the value of the anisotropy
URADMEAN: The projection of U onto a radial vector from the local origin (see above) to the atomic position.
UTANGMEAN: The average value of U projected on to a plane perpendicular to the radial vector.
UISOMEAN2, ANISOMEAN2, PROLMEAN2: If TLS groups have been fitted, then the values of Uiso, A and PROLMEAN as derived from the fitted TLS parameters are also given.
Radial distribution of Urad and Utang: In a separate graph, Urad and Utang are plotted against R**2. For rigid body motion, Urad should be constant, and Utang linear in R**2.

All quantities are averaged over the atoms in each residue, unless the MAINCHAIN is given, in which case the average is over main chain atoms only.

EXAMPLES

Runnable example

anisoanl.exam

Example 1

Example of fitting TLS groups, and plotting anisotropy etc.


anisoanl xyzin holo_adh.pdb tlsin holo_adh.tls \
  xyzout holo_resid.pdb tlsout holo_out.tls <<eof
FITTLS
RIGIDBODY OFF
PLOT
MAINCHAIN
END
eof

Example 2

Example of the RIGIDBODY analysis:


anisoanl xyzin 1exr.pdb tlsin 1exr.tls <<EOF
FITTLS OFF
RIGIDBODY
DUBINS 8 10
DURANGE 0.2
PLOT OFF
END
EOF

where the rigid groups are defined in 1exr.tls as:


REFMAC

TLS    Chain A                                                                      
RANGE  'A  16.' 'A  16.' CG CD1 CD2 CE1 CE2 CZ                       

TLS    Chain A                                                                      
RANGE  'A  19.' 'A  19.' CG CD1 CD2 CE1 CE2 CZ

The two rigid groups are two PHE side chains. This example gives a clear indication of the two phenyl groups acting as rigid-bodies. Similar results can be obtained for domain-size quasi-rigid bodies, though never as clear-cut.

In this case, the postscript plot is unhelpful - it is more helpful for looking at larger groups of atoms.

Published applications of ANISOANL

arginine kinase transition-state analogue complex at 1.2A: Yousef M.S., Fabiola F., Gattis J.L., Somasundarama T. and Chapman M.S. (2002) Acta.Cryst., D58, 2009
calmodulin: Wilson, M.A. and Brunger, A.T. (2003) Acta.Cryst., D59, 1782
GroEL: C Chaudhry, A L Horwich, A T Brunger and P D Adams (2004) J. Mol. Biol., 342, 229

REFERENCES

Martyn Winn, CCP4 Newsletter March 2001, 39
ANISOANL - analysing anisotropic displacement parameters
R.E.Rosenfield, K.N.Trueblood and J.D.Dunitz, Acta Cryst, A34, 828 - 829 (1978)
Rigid-body postulate.
T.R.Schneider, Proc. CCP4 Study Weekend, 133 - 144 (1996).
Application of rigid-body postulate to protein SP445.
V. Schomaker and K.N.Trueblood, Acta Cryst., B24, 63 - 76 (1968)
Original description of TLS.
V. Schomaker and K.N.Trueblood, Acta Cryst., B54, 507 - 514 (1998)
Description of THMA program for small molecules, which fits TLS parameters (and more) to refined U values.

AUTHORS

Martyn Winn (m.d.winn@ccp4.ac.uk)