[gmx-users] Pulling a CG protein

[David van der Spoel]

Marc Baaden wrote:
> Hi Xavier (and Johnny),
> 
> I quite agree with what Xavier says. Still I would like to point out
> that we have used CG models to pull on them and at least qualitatively
> they behave quite reasonably, although these models have never been
> parameterized or systematically tested with this kind of application in
> mind.
> 
In general models that have been parameterized for one specific 
observable may not reproduce other observables *unless* the observables 
used for parameterization have a clear physical basis *and* the energy 
function captures the correct physics. Most biomolecular force fields 
(including Martini as far as I know) have been parameterized to 
reproduce density and energy of condensed phase (liquids), and hence 
this kind of data is usually reproduced quite well. However mixtures of 
components (e.g. two liquids) are usually not reproduced as well.

By doing the kind of pulling simulations, preferably both atomistic and 
with CG models, one can try to verify whether the CG models can actually 
reproduce force curves to a comparable accuracy as all atom models. This 
kind of work should be done *systematically* before any applications.

What does this boil down to? If you want to apply MD tools to get an 
accurate force curve *now*, use all atom models. If you want to play 
with and test CG models, then do a large systematic test (many systems). 
If the CG models perform reasonably, you have cleared everyone's 
conscience and will get a lot of citations. If they don't then you 
should try to understand why not. Are there fundamental flaws in the 
model? If so, can they be fixed?

Have fun.

> What I mean by qualitatively is that I wouldn't trust a force-extension
> curve of such simulations (unless you do some force-matching :)) - re:
> FD144), but the conformational changes that are observed in a moderate
> pulling regime do seem to make sense.
> 
> The experiments we carried out were interactive pulling experiments,
> either to induce opening and closure of the guanylate kinase enzyme or
> to explore the membrane anchoring and the mechanical properties of the
> SNARE complex. This is briefly described in reference [1].
> 
> Cheers,
> Marc
> 
> 
> [1] http://dx.doi.org/10.1002/jcc.21235
>     O. Delalande, N. Férey, G. Grasseau and M. Baaden:
>     Complex Molecular Assemblies at hand via Interactive Simulations,
>     2009, J. Comp. Chem.
> 
> 
> x.periole at rug.nl said:
>>> Hi Johnny,
>>> I am not familiar with pulling and even less with gromacs but I would
>>> be very cautious in using the MARTINI force field for the kind of
>>> simulation you are doing. This CG model has not been tested at all
>>> for this and it might not be very good at it! But I would be very
>>> interested in knowing how it   actually perform.
> 
>>> XAvier.
> 
> 
> 
> 


-- 
David van der Spoel, Ph.D., Professor of Biology
Molec. Biophys. group, Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205. Fax: +4618511755.
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://folding.bmc.uu.se