On Thu, Jun 13, 2013 at 11:21 PM, kurisaki <
kurisaki.ncube.human.nagoya-u.ac.jp> wrote:
> Dear Jason,
>
> Thank you for your kind support and useful advice.
>
> I solve the problem by using your excellent program, "parmed".
> It help me to achieve the aim, more quickly and accurately.
> I can perform the simulations I intended.
>
> By the way, I have another problem in terms of SHAKE when I performed
> simulations
> With the modified prmtop.
> As you know, I set pseudo bonds between atoms in protein and ions.
> Then, ions bonds with hydrogen, too. These cause anomalous geometry,
> thus stopping simulation.
>
> To avoid such a problem,
> I edited "FLAG BONDS_INC_HYDROGEN" and "FLAG BONDS_WITHOUT_HYDROGEN",
> where ion-hydrogen bond pairs were moved from the former to the later.
> However, it does not work; sander cannot access these additionally modified
> prmtop file.
> In fact, simulations were stopped by showing
> | Note: 1-4 VDW scale factors were NOT found in the topology file.
> | Using default value of 2.0.
> at the last line of mdout.
>
> Additionally, "ambpdb" did not work, too.
> Then I guess such an additional modification break a PRMTOP format...
>
It doesn't 'break' the prmtop, but it breaks certain assumptions that many
programs may make. Modifying the prmtop is a powerful way of testing out
certain ideas, but it's potentially dangerous as well... While ParmEd
mitigates the danger to some extent by automating the 'mundane' details
like making sure all of the pointers are up-to-date and consistent, it
doesn't necessarily stop you from making a change that results in a
completely meaningless simulation.
If it cannot be solved easily,
> I set SHAKE off in simulations.
> But If you have some idea to solve this problem,
> I am most grateful to hear that.
>
I think you will have to turn off SHAKE. The problem with keeping it on is
that the equilibrium distances will not be maintained between the ions and
the hydrogen, since the bond force constant is zero (and the bond is just
present to trigger the exclusion of that pair).
Another thing you may not have anticipated: what you are doing is only
eliminating the direct-space nonbonded interactions. The electrostatic
interactions between each ion and the periodic images of the protein are
still 'active' in the reciprocal space part of the Ewald sum. There is no
way of turning this off without changing code since the Ewald sum is
inherently non-local.
HTH,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 13 2013 - 21:00:02 PDT
