Re: [AMBER] Entropy Decomposition

From: Thomas Cheatham III <>
Date: Fri, 24 Jun 2011 15:14:19 -0600 (Mountain Daylight Time)

> If we use quasi-harmonic approximations instead of normal mode analysis,
> then the decomposition is possible. Or not?

You *can* calculate from normal modes or quasi-harmonic the vibrational
entropy for each isolated amino acid, however this will not tell you much
since (a) this result is not very sensitive to changes in the amino acid
structure, (b) it neglects the coupling of the surroundings. You can try
this to verify to yourself...

As Professor Case mentioned, the dominant entropic factors are from the
low-frequency modes which represent large-scale collective motions among
many atoms, for example bending or opening. At the opposite extreme are
the high frequency motions (bonds) which are only between two atoms--
these are relatively independent of the environment and contribute very
little (or effectively a constant amount) to the entropy.

90% of the motion (and therefore the conformational entropy around a given
substate/minima) is in the first ~10% of the modes. By decoupling to the
residue level, you effectively loose this.

If I were to think about decoupling, I would look at means to estimate
configurational entropy (changes) of the molecules rather than
vibrational; i.e. are particular rotameric states locked out comparing
free vs. bound? Is there greater freedom of movement in one simulation
compared to another? Estimating configurational entropy differences is
more tricky; I would look for work by MK Gilson and co-workers.


AMBER mailing list
Received on Fri Jun 24 2011 - 14:30:04 PDT
Custom Search