Re: [AMBER] cis form: avoiding atom overlap in all-atom modeling

From: bio lab <biolabcs.gmail.com>
Date: Fri, 19 Oct 2012 19:11:16 +0200

> The cis/trans states are controlled by a double-bond,
> so that should mean that the only way to change between them
> is for that double-bond to temporarily break no?

Yes, exactly. This is what happens in reality: the molecule starts in trans
conformation, sunlight (or UV light) temporarily breaks the double bond,
the dihedral angle rotates, and finally the bond reforms when the molecule
is in cis conformation.

> Unless that isn't treated as a full double-bond and you modelled it
> as some kind of peptide-bond like thing?

Again, correct. I model it with a dihedral term with multiplicity 2 and a
large force constant, similarly to the omega angle of a peptide bond.

I think this is possible with MD, no?! Forget for a moment my specific
case, and let’s imagine a simpler situation, in which there are no steric
problems. For instance, the chain of a fatty acid with a single
(unsaturated) double bond in the middle. I think one can use MD to model a
cis-trans transition... this is unlikely to happen on a short time scale,
but it could happen in MD at very high (although unrealistic) temperature.
Please correct me if I am wrong on this.

Now, back to my case. I agree that charges are not perfect and also the
force constant should probably be adjusted, and this is what I was expected
to work on. But then I stopped when I realized that the dihedral never ends
up in cis (0°) and always returns to trans (180°) even when it has crossed
the energy barrier (<90°). This is what I see happening at high-T. So I
started to worry that charges and force constant would be only “premature
optimization”, and there is a deeper problem with the modeling of the
structure in the cis form, due to steric clash of atoms.

Please notice that I still see this more as a problem of “static” modeling
of the stable cis conformation, rather than of a correct dynamics of the
transition between the two rotamers. I might be wrong on this, too.

Thanks again for your attention.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Oct 19 2012 - 10:30:04 PDT
Custom Search