Hi Xie,
> Many thanks to all who responded to my earlier query (Drs. Paul
> Swepston, Randy Read and Nicholas Glykos). I am trying to determine the
> structure of a very long coiled coil dimer (roughly 150 residues long)
> by molecular replacement. I don't know if it forms a canonical coiled
> coil from end to end (it probably doesn't). I seem to get the right
> solution using different lengths of polyala (and polygly and modeled
> ones with side-chains) coiled coils as my search models (maps show some
> density for unmodeled parts of the coiled coil structure).
If your current best phase-set gives you back correct information that was
not part of your model, you are doing fine. You are getting there (but it
may take a while).
Starting from your current best polyAla model, you can give it a try with
very many runs of torsion angle simulated annealing (starting from a high
temperature), or with rigid-body simulated annealing (both from within CNS
or XPLOR). Don't do positional refinement. With an almost perfect polyAla
model you should expect R-free and R in the low 40s (%) range for ~2A
data.
> However, building a model accounting for the entire sequence, getting
> the correct chain direction and placing residue side-chains have proved
> difficult. And my R-facs have never improved beyond the low 50s during
> different refinement attempts. At the moment, I have 3 more or less
> contiguous 30 residue stretches of coiled coil found by 3 independent
> molecular replacement attempts using epmr (This was done to account for
> possible bends, stutters in the dimer). I am trying to get the rest of
> the model by new rounds of molecular replacement using epmr. I have
> carried out exhaustive runs using epmr. But I haven't found the rest of
> the model. What are the options available to me with molecular
> replacement? I have tried molrep, phaser and amore from the ccp4 suite.
> My diffraction data is quite good (data between 25 and 2 A with an Rsym
> of 8%).
I'm not sure whether what you really need is more molecular replacement.
Maybe you should invest more time with the maps (combined with 'safe'
simulated-annealing-based optimisation methods). What I've just said only
applies if these 90 residues you have can indeed produce correct
information that was not part of the initial model (as you said).
> Can I run Qs when a partial structure is available?
No. Sorry.
Nicholas
--
Dr Nicholas M. Glykos, Department of Molecular
Biology and Genetics, Democritus University of Thrace,
University Campus, 68100 Alexandroupolis, Greece, Fax +302551030613
Tel ++302551030620 (77620), http://www.mbg.duth.gr/~glykos/
