AMBER Archive (2009)

Subject: RE: [AMBER] Comparison of the CHARMM and AMBER parameters for alkane

From: FyD (fyd_at_q4md-forcefieldtools.org)
Date: Fri Dec 04 2009 - 11:44:57 CST


Hello Stephane,

> My question came because I have done some MD with the CHARMM (Klauda et al.)
> and GLYCAM06 ff and I would like to explain in my paper why I found some a
> slight difference (~12%, GLYCAM06=84% and CHARMM=72%) in the average trans
> population for the CCCC angles in the alkyl chain between these two ff.
> Since these results are similar that others ones obtained in my lab with the
> FF99SB parameters, I am not worried about this result.

May-be the differences underlined by Ross are responsible of these 12
% differences...

However: Did you repeat the simulations you made using the GLYCAM06
conditions (for instance) & using different initial conditions ? What
is the error/uncertainty observed using a single FF and different
initial conditions ?

Indeed, starting from different initial conditions and using the same
FF, a 10 % uncertainty for a structural feature is quite common from
what we observed... However, my feeling is that this uncertainty (or
error) in MD simulations is related to the total time of the
simulation and to the observed parameter itself.

What would be the expected uncertainty value for a parameter observed
during MD simulation when using a single FF condition and different
initial conditions ?
Thus, I am not sure a 12 % difference between two different force
fields is representative of something...

I am sure people in the list have answers for that...

regards, Francois

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber