On Wed, Jul 19, 2017, Ramin Salimi wrote:
>
> To calculate the configurational entropy, I used quasi-harmonic
> analysis. However, the entropy for all the elongations showed pretty
> much the same number, only differing in the digits after decimal.
I think we would need details about how you did the quasiharmonic analysis:
how many snapshots did you use? (And how big is your system)?
>
> My guess is that it is calculating the entropy of the constraining
> force, because in all the production runs corresponding to different
> lengths, I have employed the same input files with the restraining
> constant k=100 to keep the molecule in that particular extension.
Quasiharmonic analysis just looks at the fluctuations about the average
structure, and just uses coordinates and masses as inputs. It has no idea
about whether you have restraints or not.
>
>
> My second question is why do I see translational, vibrational,
> rotational entropy contributions listed in the thermo.dat file along
> with the total entropy while I have already did rms to the first frame,
> find the average, and then rms to the average structure to REMOVE THE
> GLOBAL TRANSLATIONS AND ROTATIONS. I was expecting only vibrational
> entropy???
The thermo.dat analysis calculates translational and vibrational entropies
using the rigid-rotor model (using the average structure). So the
translational entropy should be the same for all extenstions, since you have
the same molecule. You may choose to ignore the rotational entropy if you
wish.
>
>
> My third question is when we are trying to find the entropy, the number
> of eigenvectors in front of "vecs" in the diagmatrix command should be
> the same as the number of atoms in the mask whose average structure
> was computed a few steps earlier. Is that right? Because my intuition
> says to find the entropy, we should take into account all the degrees
> of freedom except hydrogen atoms for example in a DNA molecule, or only
> backbone atoms in protein systems?
The only straightforward method to relate quasiharmonic calculations to
thermodynamics is if you choose all atoms. There are some empirical relations
that relate calculations on subsets of atoms to the full calculation, but they
are unlikely (in my view, anyway) to work for things like stretched DNA.
First thing is to figure out why you keep getting the same answer for
different geometries of the DNA. Be aware that you will need *lots* of
sampling and long trajectories to get converged results.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jul 19 2017 - 06:00:11 PDT
