[gmx-users] Re:

[Erik Lindahl]

>

Hi Philippe,

The best place for this type of questions is to subscribe & post to the gmx-users
mailing
list - you'll probably get a better & faster answer, and once a question has been

solved other users can find it in the archives. I'm cc:ing this one there...

> Hello,
>
> I'm a french student using Gromacs 2.0, and I've a problem with distance
> restrain between 2 atoms.
> I am in the case I need in my simulation that 2 atoms must stay close to each
> other at a distance of 2 Angstrom, with a strenght equal to a covalent bond...
> Actually, I proceed in this way:
> I modify the .top file before the last RUN (ie after minimisation steps) by
> writing the lines:
>
> #ifdef DISRES
> ;distance restraint
> [ distance_restraints ]
> ;ai     aj      type    index   type'   low     up1     up2     fac
> 899     1250    1       0       1       0.2     0.21    0.22    1.0
> #endif
> (where 899 and 1250 are the reference number of my two atoms)
>  and in my mdp file, I use the NMR refinment with options :
>
> disre=simple
> disre_weighting=equal
> disre_mixed=no
> disre_fc=1000
> disre_Tau=0
> nstdisreout=1000
>
> My questions are :
> Do I modify any of this? Do I use a disre_Tau <> 0?

The weighting and simple/ensemble options only matter when you have more than
one restraint and want to average, so that's ok.

fc is the force constant - 1000 is default, but if you really want something as
hard
as a bond (that's HARD) you can have a look in the forcefields, e.g. the file
ffgmxbon.tip - the "kb" column in the bondtypes section is the force constants.

The "tau" option turns on running average if it is larger than zero - that is we
first
calculate the running average of the bond length over tau ps, and then apply the
force depending on the average length.


>
> With my options, how can fluctuate (in length) the bend at a Temperature of
> 360K?

That is not completely trivial to answer - in principle it can be derived by
statistical
mechanics; the fluctuations in energy will be of the order kT where k is
Bolzmanns
constant and T the temperature, and since the energy of the restraint is

V=0.5*kb*(r-r0)^2

the length deviation should be of the order

r-r0=sqrt(2kT/kb)


This might not be a very good measure in practice, though, since there might be a
lot
of other forces trying to separate your restrained atoms. The best option is
usually
to run a simulation and check the output in the log file!

Cheers,

Erik