Since others may benefit from an archive of this discussion, please keep
responses to the mailing list unless we request private communications
(typically to receive test files). I'm also not the best person to ask
for all of these questions.
On Thu, 2014-04-17 at 18:44 +0530, MOHD HOMAIDUR RAHMAN wrote:
> Dear Swails
>
>
> Sorry for contacting in personal mail.
>
>
> I have already run MD with full Ewald electrostatics calculation.
I was not referring to the original calculation; I was referring to the
interaction energies you were computing.
>
> For to see cutt-off effect in Average Interaction Energy calculation.
> In a separate test of pairwise calculation, I used the default value
> without any cutt-off. like below command.
There is no such thing as "no cutoff" in a periodic system. Since there
are an infinite number of particles, you still have a cutoff (typically
dictated by the box size and the minimum image convention (MIC)). Since
unit cells are never spherical, the "MIC cutoff" is a little weird.
>
>
> .........................................................................................................
> trajin urea_wat.crd
>
>
> pairwise U_U :URA out pairwise_uu.dat
> pairwise S_S :SPC out pairwise_ww.dat
> pairwise U_S :URA,SPC out pairwise_us.dat
> ...........................................................................................................
>
>
> The out put value gives exactly the same value as compare with 12.0
> cutt off value. So may be my cutoff is not playing major role here,
> because giving same value in both case with and without cutt off
> (other case it will, my system is only 100 Urea molecules and 400
> water molecules, so this is very small system).
>
>
> Here I am right ?
Not necessarily. As I said above, you still have a cutoff. Also, the
"pairwise" action does not use periodic boundries, which can cause
errors as well (_especially_ if you do not image before running the
calculations). Even if you do, though, the fact that you are computing
interaction energies between species dispersed throughout the box means
that you are also including interactions for particles near the edge of
the box. These particles will "feel" a vacuum beyond the edge of the
box when it should not.
In light of this, "pairwise" may not be the best way to compute these
"interaction energies". NAMD has a "pairInteraction" keyword that will
compute interaction energies with full PME (using a difference of 3 PME
calculations).
Also don't forget that the only thing that matters in thermodynamics is
that the _total_ free energy of the system must be minimized. Consider
a DNA duplex; each strand strongly repels each other (they're both carry
a large negative charge), but their interactions with the solvent and
ions stabilizes the duplex more than the phosphate-phosphate
interactions between strands destabilize it. A similar thing may be
happening here -- the water-urea interactions may be stabilizing the
otherwise-unfavorable arrangements of urea molecules. Since there are 2
amine groups and only one carbonyl on a urea molecule, there are
probably more 'unfavorable' relative orientations of 2 urea molecules
than there are favorable ones.
>
> Also I understand all that you suggested me. Please elaborate this
> sentence more so I will get more clear information from this.
> Are you sure you don't want some other metric, like maybe a radial
> distribution function (pair correlation function) that you could use
> to
> back out a free energy?
A radial distribution function is just a probability distribution
function (it's the probability of finding a particular atom as a
function of the distance away from a particular species). As you should
know from basic statistical thermodynamics, probabilities are directly
related to free energies. You can use this to compute a PMF from the
RDF.
I think what you need here, though, is to sit down with a pen and paper
and write down exactly what you are trying to calculate, what those
numbers actually _mean_, and then come up with a strategy for how to
compute those quantities.
>
> Thank you in advanced for giving me yours valuable time.
HTH,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Apr 17 2014 - 07:30:03 PDT
