It is our intention that it should be easy for the user to apply LSSR similarity restraints.
Accordingly command line options are provided for the most common applications. For expert users and sophisticated treatments
.Gelly cards can be used. The process of switching from the automated options to a .Gelly card treatment is described below.
The .Gelly cards themselves are described in the next appendix.
The -autoncs command line option will attempt to identify NCS within a structure and apply
LSSR similarity restraints on positions. The -autoncs command line option
can be used on the BUSTER command line or the gelly_refine command line,
The identification compares all pairs of chains within in a structure. For each pair it will assess
the proportion of residues that has the same residue number and residue type (e.g., "ALA"). If this is
above 80% then NOTE BUSTER type cards are written to activate LSSR restraints for the NCS identified.
Note that water molecules are ignored by -autoncs. If no NCS is identified the -autoncs
will report this and the run will continue.
Note that no restraint is placed on the temperature factors by the -autoncs option. If you wish to
place harmonic restraints on temperature factors then this can be done by using the
NOTE BUSTER_SIM_RESTRAIN_B.Gelly card.
The process of switching from the automated options to a .Gelly card treatment is described below.
A practical example demonstrating the use of the -autoncs
can be found in the 002 run of the
basic NCS setup
example found at the
Global Phasing BUSTER Wiki.
LSSR restraints remove much of the need to "prune out" parts of the structure that are
NCS violators. This is because markedly distinct parts of the structure will result in different interatomic
contact distances (so LSSR restraints will plateau). A problem arises where there are distinct but
rather similar side chain conformations in different NCS copies.
Here LSSR restraints will work hard to pull the conformations together. The -autoncs procedure includes
a simple routine designed to identify such regions and automatically prune them from the LSSR restraints applied. The
procedure is conservative taking the attitude "if in doubt prune it out". It will therefore remove some parts of the
structure that could happily be left in NCS restraints. It is also not perfect and you may need to manually adjust
the pruning, as described below.
If you want the automated NCS but do want the automated pruning option to used then use the -autoncs_noprune
option rather than the -autoncs option. This can
be a good idea at the start of a refinement where it can be expected that minor differences between NCS copies
are yet to be of importance. The
basic NCS setup
example found at the
Global Phasing BUSTER Wiki.
shows how the -autoncs_noprune option can be usefully used (003) following on from an initial run with
-autoncs_noprune. The pruning is shown to result in both a slight improvement in Rfree as well as
getting rid of some difference map features caused by LSSR pulling together residues that are conformationally close but distinct
in the different NCS copies.
The default weight for autoncs LSSR restraints is 2.0/(Number_of_chain_in_NCS-1).
This means that 2 fold NCS is weighted 2.0, 3 fold 1.0, 4 fold 0.66, 5 fold 0.5, 6 fold 0.40. The value used is reported to
LIST.html.. It should be noted that each NCS relation between two chains creates a separate set of LSSR restraints.
So this automatic weighting compensates for this and ensures that high order NCS is not overweighted.
To change the weight
for -autoncs (or -autoncs_noprune) specify the -autoncs_weight value option. Note
that the value is absolute. So to if you have 6 fold ncs and you want to double the default weight specify -autoncs_weight 0.80.
Target restraints enable the exploitation of the similarity of
the structure under refinement to an already known structure.
The command line option -target filename.pdb sets up basic
LSSR similarity restraints to a target structure.
The -target option will set up a similarity group matching all atoms in the refined structure
to all atoms in the target structure. This means that the chain identifiers (and residue numbers) must correspond between
the two. If the chain ids do not match then you should be able to use a -Gelly file option instead of
the -target option.
See Specifying target type restraints with gelly cards for help with this.
It is not necessary or desirable to displace the coordinates into a common frame to the structure being refined.
LSSR restraints are based on close interatomic distances with a structure so superposition is irrelevant.
it is common that two different NCS copies can be very similar other than atoms
CD1 and CE1 match CD2 and CE2 and vice versa. The -sim_swap_equiv
option works through the residues in the molecule checking whether the NCS (and/or
target) restraint function is improved by swapping the coordinates, temperature factors
and occupancies of relevant pairs of atoms. The check is done at the beginning of every big cycle
and details of any residues swapped are written to the relevant LIST.html file.
The use of this procedure will mean that IUPAB naming conventions for sidechains are
likely to be broken. It is intended to address this in a future release.
The -sim_swap_equiv_plus option extends the swapping to residues that
are quasi-symmetric such as ASN, GLN and HIS. Using this option may well disrupt hydrogen bonding
patterns and it is advised to check its result. The option is dumb - if a residue has good hydrogen
bonding in one copy but bad in the other it is equally likely the good will be swapped to the bad!
In general, for a given refinement the procedure needs only be applied once and should
then not be used (it is slow and could disrupt sensible hydrogen bonding patterns).
The
basic NCS setup
example found at the
Global Phasing BUSTER Wiki shows how
the -sim_swap_equiv_plus option swaps 65 residues to improve NCS. This results in improved
Rfree, Rfree-Rwork gap and geometry scores compared to leaving equivalent atoms to remain different.
Note that -sim_swap_equiv and -sim_swap_equiv_plus procedures will
work both for LSSR restraints, superposition-based (RMSD) and/or temperature factor restraints.
The restraints can be produced by command line options or in a .Gelly type file.
The -autoncs and -target options produce basic
LSSR similarity restraints. In the final stages of refinement it is
likely that you will need to adapt the restraint conditions for your needs.
The first thing to do is locate the LIST.html output for the final cycle of the refinement
made with the -autoncs and/or -target options. The location of this file
will be listed in the "top" log written by BUSTER to standard output. In general it will something like:
foo/01-BUSTER/Cycle-5/LIST.html where foo is the directory specified in the
BUSTER -d option. Open this file with a web browser (firefox). You will find log output from
the automated restaint options that should include lines with a hyperlink to the file auto_gelly_cards.txt
-autoncs option .Gelly cards also written to file auto_gelly_cards.txt
The file auto_gelly_cards.txt contains the .Gelly cards produce by the the -autoncs and/or -target options. (You could also find it with the linux find command).
Copy the contents of this file to a file - lets call it manual_gelly_cards.txt
Use your favourite editor to edit manual_gelly_cards.txt to reflect what you want.
To do this you will need to understand the .Gelly cards used.
These are described in the next appendix.
For the next run drop the -autoncs and/or -target options. Instead specify
-Gelly manual_gelly_cards.txt
Page Author: Oliver S. Smart
Please send feedback to: buster-develop@globalphasing.com
Last modification: 24.11.2016