COMPARE -- compare 3D structures given alignment

ALIGN_CODES = <string:0>	'all'	codes of proteins in the alignment
ATOM_FILES = <string:0>	''	complete or partial atom filenames
ATOM_FILES_DIRECTORY = <string:1>	'./'	input atom files directory list (e.g., 'dir1:dir2:dir3:./:/')
OUTPUT = <string:1>	'LONG'	selects output: 'SHORT' | 'LONG' | 'RMS' | 'DRMS'
MATRIX_FILE = <string:1>	'family.mat'	the filename of the pairwise distance matrix
COMPARE_MODE = <integer:1>	3	selects the type of comparison: 1 | 2 | 3
RMS_CUTOFFS = <real:11>	3.5 3.5 60 60 15 60 60 60 60 60 60	cutoffs for RMS, DRMS, Alpha Phi Psi Omega chi1 chi2 chi3 chi4 chi5
FIT_ATOMS = <string:1>	'CA'	whether to superpose before comparison
DISTANCE_ATOMS = <string:2>	'CA' 'CA'	atom type used for variability calculations
FIT = <logical:1>	on	whether to do pairwise least-squares fitting or ALIGN2D alignment
ASGL_OUTPUT = <logical:1>	off	whether to write output for ASGL

Description:

This command compares the structures in the given alignment. It does not make an alignment, but it calculates the RMS and DRMS deviations between atomic positions and distances, and class differences between the mainchain and sidechain dihedral angles. In contrast to the SUPERPOSE command, COMPARE works with a multiple alignment and it writes more information about the pairwise comparisons.

If no alignment is available, it assumes a 1:1 correspondence for the proteins specified by ATOM_FILES or ALIGN_CODES. If ATOM_FILES is defined, it is used with the MODELLER file-naming mechanism (Section 2.1.4) to find full names for the atom files. If it is not defined, ALIGN_CODES, which is usually set by the previous READ_ALIGNMENT command, is used. ALIGN_CODES does not have to be set if ATOM_FILES is set.

OUTPUT selects short ('SHORT') or long ('LONG') form of output to the log file. If it contains word 'RMS' or 'DRMS' it also outputs the RMS or DRMS deviation matrix to file MATRIX_FILE. This file can be used with the PHYLIP program or with the DENDROGRAM or PRINCIPAL_COMPONENTS commands of MODELLER to calculate a clustering of the structures.

COMPARE_MODE selects the form of the positional variability calculated for each position along the sequence:

1,

for true RMS deviation over all proteins that have a residue at the current position. This does not make any sense for periodic quantities like dihedral angles.

2,

for the average absolute distance over all pairs of residues that have a residue at the current position.

3,

the same as 2 except that average distance, not its absolute value is used (convenient for comparison of 2 structures to get the $\pm$ sign of the changes for dihedral angles and distances).

RMS_CUTOFFS specifies cutoff values for calculation of the position, distance, and dihedral angle RMS deviations for pairwise overall comparisons. If difference between two equivalent points is larger than cutoff it is not included in the RMS sum. The order of cutoffs in this vector is: atomic position, intra-molecular distance, $\alpha$, $\Phi$, $\Psi$, $\omega$, $\chi_1$, $\chi_2$, $\chi_3$, $\chi_4$, and $\chi_5$ (there are 5 dihedrals in a disulfide bridge), where $\alpha$ is the virtual ${C}_\alpha$ dihedral angle between four consecutive ${C}_\alpha$ atoms. These cutoffs do not affect positional variability calculations.

FIT_ATOMS string specifies all the atom types (including possibly a generic 'ALL') to be fitted in the least-squares superposition. These atom types are used in the least-squares superposition, and in calculation of the position and distance RMS deviations.

DISTANCE_ATOMS[1] specifies the atom type that is used for getting the average structure and RMS deviation at each alignment position in the ASGL output file 'posdif.asgl'. This ASGL file contains the positional variability of the selected atom type in the family of compared proteins. The ASGL output files can then be used with ASGL scripts 'posdif' and 'dih' to produce POSTSCRIPT plots of the corresponding variabilities at each alignment position. ASGL_OUTPUT has to be on to obtain the ASGL output files.

If FIT = on, a least-squares superposition is done before the comparisons; otherwise, the orientation of the molecules in the input atom files is used.

Example: See MALIGN3D command.