Re: [AMBER] Problem in MD simulations of DNA using ABC protocol

From: Thomas Cheatham III <>
Date: Tue, 9 Nov 2010 11:35:29 -0700 (Mountain Standard Time)

> But after running the simulations for 5 ns, I found out that my structure is
> distorted in the sense that the rings of bases are no longer planar. Can you
> tell me where I am making the mistake or is their the problem in the
> visualizer (I am using Pymol and ViewerPro for visualizing the structures) ?

The structure looks perfectly fine to me as representative of a snapshot
from MD at 5 ns.

Why do you expect the bases to be perfectly planar? If you take a random
snapshot, thermal fluctuations will be evident as distortion away from
ideality. If you look at the force field, there are force constants to
maintain planarity, however they are not very strong as in real life the
bases (and all the atoms) are dynamic / flexible. MD = molecular
dynamics. In all atom MD, all of the atoms have independent motion. Look
at the movie and you will see... If you want ideally rigid bases, this
would involve constraints and would seriously complicate the
implementation (and be less realistic).

To get to a nearby local minima, you could minimize the structure; this
will tend towards making the bases more planar (depending on the local
environment). Also, if you are sampling about a given local minima, you
could create an average structure which would presumably (assuming you
remain within that local minima) provide a representative structure for
that minima. Of course if groups are rotating with equivalent groups (for
example the thymine methyl) the average structure will have all the
hydrogens overlapping (parallel to the methyl rather than in tetrahedral

Note that the 5'-terminal base distortion (top of PDB) suggests you may be
running without the parmbsc0 modifications; these are likely important for
DNA as otherwise most of your backbone gamma angles will go to trans as
the simulation proceeds further...

Finally, as has been mentioned previously, on the list we prefer
names/affiliations to be listed for example by signing the e-mail or
having an e-mail address that identifies the sender. Although I did
respond this time, people (and myself) are in general are less likely to
respond to an anonymous post...


AMBER mailing list
Received on Tue Nov 09 2010 - 11:00:03 PST
Custom Search