Norbert,
Sorry for this late reply. SSM will find multiple alternative matches
but most implementations just show you the 'best'. The Superposition
tool in CCP4mg uses ssm and has an option to look through the non-
optimal matches - hopefully one of these will be the one that you want.
Liz
On 17 Feb 2010, at 10:35, Norbert Straeter wrote:
Dear All,
last Friday I posted the following question:
> I have a heterodimer of two related subunits (A and B). Now I
> want to superimpose it using a secondary structure matching
> (SSM) algorithm (e.g. CCP4 superpose or the coot algorithm)
> such that the superposition matches A onto B and B onto A at
> the same time. For a perfectly symmetrical closed C2 dimer,
> it would not make a difference if I just superimposed A onto
> B, but it is unsymmetrical. I know that this is easy to
> specify in lsqkab when knowing the matching residue pairs.
> Does anybody know an SSM program where I can specify the
> chains to be superimposed (A->B and B->A)?
Some replys described how to specify selections in pymol or coot for
SSM alignment. To make it more clear, if you specify chains A+B as
the moving selection and A+B or B+A as the reference, you will
always get the best but trivial superposition of A->A and B->B. It
appears as if there is currently no program for SSM superposition
for which you can specify the chains to be matched if the structure
that needs to be superimposed contains internal symmetry in form of
different chains. As a workaround, one can rename chains and residue
numbers before and after alignment:
OK, here's a possible inelegant solution, till something better comes
up....
Take dimer 1, and put it all into one chain X, keeping numbers for
A and
adding an offset eg 500 for the B chain. Take dimer 2, and put it
all into another chain Y, keeping the numbers
for B and adding the same offset to A.
Then do a match of chain X with chain Y.
(Judith Murray-Rust)
Could it help to merge chains A and B first into two longer chains,
where
C=A+B
and
D=B+A ?
(Matthias Zebisch)
Using this strategy, you can probably use your favourite SSM
alignment program. I just tested it within coot and it works.
Another strategy is the following:
One workaround would be to let SSM define the matching residues
(should
be the same for A->B and B->A) and input those in to lsqkab.
(Herman Schreuder)
Thanks to all, who provided help to this problem, also to Christian
Roth, Gerard Kleywegt (use of "explicit" and "improve" commands in
lsqman), Yu Wai Chen, Mark A. White, Christina Bourne, Markus
Wiederstein (http://topmatch.services.came.sbg.ac.at/), Ed Pozharski
and Regina Kettering.
Norbert
