POLYPOSE (CCP4: Supported Program)
NAME
polypose - a program for superimposing many multi-domain structuresSYNOPSIS
polypose xyzin1 xyzin2 ... outdir xyzout1 xyzout2 ... foo.dat[Keyworded input]
DESCRIPTION
This program originating from R. Diamond is a program to superimpose several multi-domained structures. This is done in an optimal way by minimising the residuals between (n*(n-1)/2) pairwise comparisons. One of the molecules can be fixed and the others rotated to that orientation. Alternatively, the structures can be oriented so their longest dimensions are along X. Each domain is fitted to the equivalent domains in the other structures.EACH DOMAIN MUST CONTAIN THE SAME NUMBER OF ATOMS.
The program expects coordinates in Brookhaven format and will only read the `ATOM' cards. Care must be taken to ensure that each file has the same axis definitions (NCODE number).
Transformed coordinates can be written out with optionally an average structure. The output files will also be in PDB format with the same axis system as the input files. There are coordinates for each domain fitted, for each structure. The coordinates used for fitting the structure are given as REMARKs followed by the whole structure.
All the equations referred to in this documentation come from reference [1].
KEYWORDED INPUT
The program has been parameterised and the current defaults are: maximum number of molecules = 10, maximum number of sub-domains/domains = 60, maximum number of atoms in molecule = 1200, maximum number of atoms in a domain = 2000. The permitted data control statements in the form of keywords are listed below:
CHECK, COMBINE, END, FIX, INCLUDE, INDEPENDENT, INPUT, MAXCYCLE, OUTPUT, TERMINATE
MAXCYCLE <num>
This determines the maximum number of cycles to achieve the best fit between the structures. Note that it can be set to 1 if there are just two structures or the structures are of the same shape. (default=10)INDEPENDENT
If present this causes the residual (equation R0 #40 and #41) to be calculated from N(N-1)/2 distinct and independent orientations. If the card is not present then R0 is not calculated.INPUT [ CA | ALL ]
The program will either work with all the atoms or just the C-alpha atoms (default) in a residue. The number of atoms in a domain MUST be equal between molecules.INCLUDE <res1> TO <res2> FILE <num>
These define sub-domains within the <num>-th molecule. The ordering of the cards relates to the sub-domain number. Each molecule must contain the same number of sub-domains. If absent all the residues in the molecule are included.
e.g. INCL 1 to 10 file 4
INCL A1 to B10 file 1
COMBINE <num1> <num2> ... <numN>
This combines sub-domains into a domain (applies to all molecules). Sub-domains can be included several times which has the effect of weighting the atoms in that sub-domain. If there is no COMBINE card then each sub-domain is treated as a domain. Sub-domains not mentioned in the COMBINE cards are treated as domains. If the number of atoms exceeds the maximum permitted in a domain then an error will be given. Note that the atoms in a domain are paired off in order between molecules.e.g. if there are 6 sub-domains
COMB 1 2 3 COMB 2 4 COMB 2 6
gives
1st domain is 1 2 3 2nd domain is 2 4 3rd domain is 2 6 4th domain is 5
and
COMB 1 2 2 3
gives
1st domain 1 2 3 atoms of the second sub-domain have double weight 2nd domain 4 3rd domain 5 4th domain 6
TERMINATE <crit>
This defines when the the refinement will stop before reaching MAXCYCLE. The criterion is either when:
SUM [ SIN{DP/2}**2 ] < <crit>
DP = angular shift in orientation.
or
the reduction in rms in the cycle < <crit> * rms
Default <crit>=0.00001.
OUTPUT [ MATRIX | COORDS | AVERAGE ]
Defines what output will be given from the program. MATRIX will just output the orientation matrices. COORDS will give the transformed coordinates (.rot), the atoms used to fit the structure will be given as REMARKS in the PDB file followed by the whole structure. The calculations will be done per domain per molecule. AVERAGE will produce an average structure (.ave) based on the orientation calculated for each domain. Each option is progressive from MATRIX to AVERAGE. Thus OUTPUT AVER will effectively include the other options.CHECK
If the keyword is present then the program will check residue and atom name. The program will terminate if inconsistencies are found between molecules.FIX <num>
The orientation of the <num>-th molecule will be fixed and the other structures will be fitted to this molecule. If this card is not present then the longest axes of the molecules will be aligned along x.END
Terminate input.INPUT AND OUTPUT FILES
The input PDB files are assigned to logical names XYZIN1 XYZIN2 etc. The input keywords must use the same numbering, e.g. domains specified for 'file 1' should refer to XYZIN1, FIX 2 should refer to the molecule in XYZIN2, and so on. However, if the numbering used is not consecutive, the program will renumber the files, e.g. if XYZIN1, XYZIN2 and XYZIN4 are specified then the latter becomes structure 3.The output files are:
- a)
- optionally transformed coordinates <name>d<nn>.rot, where <nn> is the domain number. The root <name> can be assigned through the logical name XYZOUT1, XYZOUT2 etc. If the logical names are undefined then the root name of the input file is taken. If the logical OUTDIR is defined all coordinate files will be sent to this directory.
- b)
- optionally average coordinates <name>d<nn>.ave, where <name> is taken from the fixed coordinates or the first file (defined in a similar way to the transformed coordinates) and <nn> is the domain number.
EXAMPLES
polypose XYZIN1 dtk1.pdb XYZIN2 dtk2.pdb OUTDIR /scr1/acr/ XYZOUT1 jnk1 << + maxcycle 3 input all indep include 1 to 10 file 1 include 1 to 10 file 2 include 11 to 20 file 1 include 11 to 20 file 2 combine 1 2 output average check fix 1 +
The output files would be:
/scr1/acr/jnk1d01.rot /scr1/acr/dtk2d01.rot /scr1/acr/jnk1d01.ave
PRINTER OUTPUT
Rms differences are shown between all structures pairwise. Also, the summation of these differences is shown (R1 equation #42). R1 can be compared with R0 (equation #40 and #41) which represents the sum of the rms residuals over the best possible fit between two molecules calculated from all possible pairs. R0 is the lowest rms achievable but can only be recognised in practise when all the structures are of the same shape or there are only two structures.AUTHOR
R. Diamond MRC Laboratory for Molecular Biology Hills Road Cambridge CB2 2QH England