I didn't test solvateOct with large multimeric systems when developing
it, and no longer work on the code. One thing that might be interesting
to try would be to bond all the monomers before solvateOct, then if
successful delete the bonds. Long bonds would be easier to see after
solvation.
Bill
On 10/4/2012 12:24 PM, Jonathan Sheehan wrote:
> Hi all,
>
> Just to follow up on my question about solvateOct from last week
> (appended below)-
>
> I wasn't able to find an answer in the archive, and my testing with
> LEaP never gave me a truncated octahedral box that had 11 Angstroms of
> water around the protein on all sides (even off-center). Because I'm
> troubleshooting some instability in this simulation, I decided to play
> it safe, and I was able to work around this issue by using solvateBox
> instead of solvateOct.
>
> I think this might be a "corner-case" bug in LEaP, which only shows up
> when using a large multimeric system. But just in case I received no
> replies because the answer is too obvious, or I left out key
> information, or didn't pose my question well, I invite anyone to smack
> me with the clue-bat at this point. :-)
>
> Thanks,
> -Jonathan
>
> On Tue, Sep 25, 2012 at 2:39 PM, Jonathan Sheehan
> <jonathan.sheehan.gmail.com> wrote:
>> When I use this command in LEAP:
>>
>> solvateOct my_protein SPCBOX 11
>>
>> I naively expect to see at least 11 Angstroms of water between
>> my_protein and the nearest box edge. But when I double-check (by
>> measuring in PyMOL or VMD), I find some protein atoms within 1-2
>> Angstroms of the edge.
>>
>> My protein is a large hexamer with 12 (internally-bound) ATPs: 6 X 484
>> residues, ~23000 atoms before adding water. If I read the archive
>> correctly, solvateOct centers the system on only the first molecule,
>> so the hexamer could be off-center within the box (which can cause
>> imaging difficulties later, but shouldn't affect the calculation).
>>
>> But then, I would guess that the solvent distance on the far side of
>> the box (assuming it's correct to go straight across) would make up
>> for the deficit. E.g. if it's 10 Angstroms too close on one side, then
>> the other side should have the extra 10A, for a total distance of 21A
>> between protein and box edge. That did not happen in this case; the
>> distance to the opposite face is only 10 Angstroms.
>>
>> Unless there's a trick to the tessellation that I'm missing, I think
>> this may be a problem.
>>
>> I could use solvateBox instead, since the box appears to be nicely
>> centered on the protein. But it adds 66063 waters rather than 27288,
>> so I'd like to avoid that if possible.
>>
>> So, could anyone confirm that I should see 11 Angstroms on both
>> opposing sides (or at least that they should sum to 22 if it's
>> off-center?). Or, alternatively, reassure me that the periodic imaging
>> of the truncated octahedron somehow doesn't require that to be true?
>>
>> Many thanks,
>> -Jonathan
>>
>>
>> Additional Details:
>> I'm using Amber12 and AmberTools12.
>> I tried "11.0", rather than "11" without a difference.
>> I tried performing addions before vs. after solvateOct, without a difference.
>> I'm still searching the many posts in the archive on related topics-
>> forgive me if I've missed the answer there.
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Oct 06 2012 - 16:30:04 PDT
