Dr. Case
Each simulation at particular length was set for 4ns, containing 2000 snapshots. You pointed on a very right direction. When I look at the trajectories, What's odd to me is they vibrate very very small; I was expecting to see bubbles after a very short time, in other words, if it is initially in S-form DNA structure restrained at for example 50 A0, then it should develop a hole or "denaturation bubble" after a very short time, and the number of holes, to be viewed, to increase with length. I was expecting to view this but didn't. They all vibrate very very locally in their positions, that's it. My take was the restraining constant is too big so it doesn't let them vibrate. In fact, the simulations have a tendency to be unstable right at the transition point, so I thought that I should have treated the DNA gently by lowering the restraining force constant from k=100(I didn't try that though).
On how to calculate the entropy, that would be AWESOME if someone could help with that. What I was expecting, planning, to do was to get a matrix as the multiplication of mass and fluctuations matrices, then multiply it by some constant, then take the determinant, and then take the log of the determinant to get to the entropy. But from that file, evecs.dat, which supposedly carries the diagonalized matrix, how can someone add some other matrix to it or multiply it by some numbers, or how to take the log of that? When things get automated, your hands are tight. Simply put, given that diagonalized matrix, evecs.dat, somebody should be able to do algebraic operations on it, like multiplication by numbers, taking the log, or the determinant, etc. But how should I do that?
Kind Regards
Ramin
________________________________
From: David A Case <david.case.rutgers.edu>
Sent: Thursday, July 20, 2017 6:43:40 AM
To: Ramin Salimi
Subject: Re: [AMBER] problem Finding entropy
On Thu, Jul 20, 2017, Ramin Salimi wrote:
>
> Here is a sample input file I feed into cpptraj, where 49.0 means the
> DNA molecule at length 49.0 A0, and is the same for all the other
> lengths:
>
> parm strip.49.0.mbondi3.prmtop
> trajin strippedtraj.49.0.crd
> rms first :1-24&!.H=
> average crdset 49.0.average
> createcrd 49.0.nc
> run
> crdaction 49.0.nc rms ref 49.0.average :1-24&!.H=
> crdaction 49.0.nc matrix mwcovar name 49.0.mwcvmat :1-24&!.H= out 49.0.mwcvmat.dat
> runanalysis diagmatrix 49.0.mwcvmat out 49.0.evecs.dat vecs 20 name
> 49.0.myEvecs nmwiz nmwizvecs 20 nmwizfile dna.49.0.nmd nmwizmask
> :1-24&!.H= thermo outthermo thermo.49.0.dat temp 300.0
A key question is how many snapshots are in the trajectory? And how long
a simulation did you run? Have you viusalized the trajectories? Are the
extent of fluctuations visible different for different lengths?
>
> question: In the manual, it says that the thermo analysis calculates
> the entropy, and heat capacity using standard statistical mechanical
> formulas for an ideal gas. why should it be applicable to dna or protein
> systems? I mean how does this method find the entropy?
In an implicit solvent model, you have an "ideal gas" of DNA molecules: there
are (assumed to be) no interactions between different DNA strands.
>
> I was just expecting a diagonalized matrix which is the matrix
> multiplication of mass, and fluctuations, and openable then I would
> simply take its determinant. And finally multiply it by some other
> coefficients to get to the entropy based on the mathematical formulation
> I prefer.
The determinant is just the product of the eigenvectors. Since entropy is
proportional to the log of the eigenvectors, the thermodynamics analysis code
is just separating the total into a term for each mode: the sum of all of
these (the total vibrational entropy) is what you want. I haven't run
quasiharmonic analysis with the current version of cpptraj, so maybe someone
else on the list can show how to get the eigenvalues, if you wish to do the
calculation yourself (which is a good idea.)
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jul 20 2017 - 06:00:04 PDT
