Refinement with rigid-base TLS parameterization has been previously
explored:
Holbrook, Dickerson, Kim (1985) Acta Cryst B41, 255-262.
(the photocopy is located in the pile of dust that I maintain adjacent
to my desk)
Ethan Merritt wrote:
On Thursday 15 July 2010, Huw Jenkins wrote:
Hi,
I am currently refining some reasonably high (1.4-1.6 Å) resolution protein:RNA complex structures and was trying the approach described in Schwartz et al. Nat. Struct. Biol. 8 (2001) 761-765 where they divided each nucleotide into three TLS groups – the ribose, the phosphorus atom plus both nonesterified oxygens and the base.
Aside from issues of describing the groups at all, I would not recommend
this as a refinement protocol. The number of atoms in each group is too
small for TLS refinement to be well-behaved in the absence of additional
restraints that the current generation of programs (refmac, phenix.refine)
does not implement.
It is in principle an attractive idea to treat at least the planar bases
as individual TLS groups, but I have not had the time or opportunity to
explore this in practice. Ideally, one might like to try constraining the
group so that the glycosyl bond defined the libration axis (see for
instance the section on ADP representations in the Newsletter that Pavel
Afonine posted recently). Unfortunately again this is not implemented in
current refinement programs.
My gut feeling is that the best TLS description would be each base
(or base pair) in its own group, the use TLSMD to analyse and assign groups
for the backbone atoms. But again I have no actual experience with this,
so it's only a suggestion.
Ethan
Just as an aside:
The B factors in the PDB file 1j75, which I think corresponds to the paper
you cite, are very far from reasonable. I would not use this structure as
an example of success in choosing a refinement protocol for ADPs. Although
I also note that it is possible the B factors archived in the PDB file have
become mangled during deposition of the output from a non-standard protocol.
When I define a tlsin file that does this with the ribose, P OP1 and OP2, and
base from 3 successive nucleotides everything seems to work fine so the groups
are defined correctly. However when I divide a single nucleotide into 3 TLS
groups only the first group has its origin calculated and actually has TLS
parameters refined:
TLS origin for group 1 -3.1622999 -6.9280601 -9.6330261
TLS origin for group 2 0.0000000 0.0000000 0.0000000
TLS origin for group 3 0.0000000 0.0000000 0.0000000
TLS group 1:
T tensor ( 1) = 0.046 0.115 0.103 -0.003 -0.008 0.087
L tensor ( 1) = 1.341 1.887 5.684 0.509 -2.040 -0.436
S tensor ( 1) = 0.015 0.015 -0.257 -0.220 -0.024 0.071 0.095
0.060
TLS group 2:
T tensor ( 2) = 0.031 0.031 0.031 0.000 0.000 0.000
L tensor ( 2) = 0.000 0.000 0.000 0.000 0.000 0.000
S tensor ( 2) = 0.000 0.000 0.000 0.000 0.000 0.000 0.000
0.000
TLS group 3:
T tensor ( 3) = 0.031 0.031 0.031 0.000 0.000 0.000
L tensor ( 3) = 0.000 0.000 0.000 0.000 0.000 0.000
S tensor ( 3) = 0.000 0.000 0.000 0.000 0.000 0.000 0.000
0.000
An excerpt from my tlsin is below:
TLS
RANGE 'B 3.' 'B 3.' P
RANGE 'B 3.' 'B 3.' O1P
RANGE 'B 3.' 'B 3.' O2P
TLS
RANGE 'B 3.' 'B 3.' C1*
RANGE 'B 3.' 'B 3.' C2*
RANGE 'B 3.' 'B 3.' C3*
RANGE 'B 3.' 'B 3.' C4*
RANGE 'B 3.' 'B 3.' C5*
RANGE 'B 3.' 'B 3.' O2*
RANGE 'B 3.' 'B 3.' O3*
RANGE 'B 3.' 'B 3.' O4*
TLS
RANGE 'B 3.' 'B 3.' N1
RANGE 'B 3.' 'B 3.' C2
RANGE 'B 3.' 'B 3.' O2
RANGE 'B 3.' 'B 3.' N3
RANGE 'B 3.' 'B 3.' C4
RANGE 'B 3.' 'B 3.' O4
RANGE 'B 3.' 'B 3.' C5
RANGE 'B 3.' 'B 3.' C6
I guess I've got the syntax of the tlsin file wrong? Any suggestions for what's
wrong would be much appreciated!
Thanks,
Huw
--
Dr Huw Jenkins
Astbury Centre for Structural Molecular Biology
University of Leeds
