In case anybody encounters a similar problem, I think I found a fix. The
'setBox UNIT vdw' in the leap.scrpt file is needed to ensure that an
appropriately sized box surrounds the molecules. This is needed in the
case one manually (or as in my case, with the help of a script)
manipulates the water molecules around.
-Amarda
***********************************
Amarda Shehu,
Graduate Student, Computer Science,
Rice University.
***********************************
On Thu, 8 Feb 2007, Amarda Shehu wrote:
>
> I have a strange problem with sander minimization which I have never had
> before. I am trying to run an NVT equilibration of a solute in explicit
> solvent. I have two distinct conformations of the solute and want to see what
> happens to each of them when equilibrating each separately in explicit
> solvent.
>
> I originally had the problem that I had a different number of water molecules
> when specifying a solvent box of the same volume for each conformation. With
> Ilyas Yildirim's help, I managed to ensure I have the same number of water
> molecules in each case. Ilyas' script is quite smart and better than manually
> removing extra waters -- it removes waters from the outer layers of the box.
>
> However, now I have a strange thing that sander hangs when I start the
> equilibration with the solute fixed. On the conformation for which I did not
> remove waters, sander is fine - it runs with no problems and I can see the
> progress in mdinfo. But for the case where I removed the extra waters, no
> mdinfo file gets written out, and the .out file is empty. I see no sign of
> progress at all, even though the equilibration scripts in each case are
> exactly the same.
>
> Here is the script I am using to keep the solute fixed first:
> ----------------------------------------------------------------
> echo "STAGE1: HOLDING THE PROTEIN FIXED AND MINIMIZING SOLVENT"
> cat <<eof > mdin
> initial minimisation solvent
> &cntrl
> imin=1,
> maxcyc=5000,
> ncyc=500,
> ntb=1,
> ntr=1,
> cut=10,
> &end
> Hold the PROTEIN fixed
> 500.0
> RES 1 18
> END
> END
> eof
> sander -O -i mdin -o frame_wat_min1.out -p frame_wat.top -c frame_wat.inpcrd
> -r
> frame_wat_min1.rst -ref frame_wat.inpcrd
> -----------------------------------------------------------------
>
> If anybody has any insight, I would greatly appreciate any pointers. I am not
> sure what is going on.
>
> -Amarda
> ***********************************
> Amarda Shehu,
> Graduate Student, Computer Science,
> Rice University.
> ***********************************
>
> On Sun, 4 Feb 2007, David A. Case wrote:
>
>> On Sun, Feb 04, 2007, Amarda Shehu wrote:
>>>
>>> These two conformations have been obtained with a newly developed method
>>> in implicit solvent (using ff03 and GB). I am preparing a manuscript for
>>> publication, but my advisor has asked me to make sure that in explicit
>>> solvent, these conformations retain both their structural integrity and
>>> their energetic difference.
>>
>> Even if you did everything you originally suggested, you would only get the
>> *potential* energy difference between the two solvated systems. But the
>> implicit solvent calculation estimates the *free* energy of the solvent.
>> The difference is the entropy effect, mentioned by Prof. Duan.
>>
>> Testing the structural integrity is straightforward, and it sounds like you
>> have already done that. Getting a comparable energetic number is more
>> involved. It sounds like you probably want to think about a thermodynamic
>> integration calculation to compute the free energy difference between the
>> two
>> solutes.
>>
>> ....good luck...dac
>>
>> -----------------------------------------------------------------------
>> The AMBER Mail Reflector
>> To post, send mail to amber.scripps.edu
>> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Feb 11 2007 - 06:07:24 PST
