AMBER Archive (2005)

Subject: Re: AMBER: How is the origin of the grid defined?(PBSA question)

From: Ray Luo (rluo_at_uci.edu)
Date: Mon Aug 22 2005 - 20:15:48 CDT


Hi Liu,

Are you using the same radii for Delphi and PBSA? Thanks a lot for
pointing out the typo in the patch file.

Best,
Ray

JunJun Liu wrote:

> Dear Dr.Luo
>
> Thank you very much! The patch works! But I don't know why
> they(results from pbsa/Delphi) are so different? Which program
> performs better? Here's my results:
>
> # DELTA
> # -----------------------
> # MEAN STD
> # =======================
> PBTOT -37.91 18.33 # pbsa, unpatched
> PBTOT -57.31 4.23 # pbsa, patched
> PBTOT -38.26 5.56 # DelPhi
>
> By the way, it seems a dot was missed in the modified code,
>
>> At line 558 in pb_force.f:
>>
>> change:
>>
>> if ( abs(totcrg) > smallcrg ) then
>>
>> to:
>>
>> if ( .false. and. abs(totcrg) > smallcrg ) then
>
> ^^^^^^
> Here I believe it should be .and.
>
> Best Regards!
>
> Liu
>
>
>
> On Tue, 16 Aug 2005 17:03:04 +0800, Ray Luo <rluo_at_uci.edu> wrote:
>
>> Dear Liu,
>>
>> Most likely this is due to the surface charge scaling for nearly
>> neutral molecules in pbsa. After removing this feature, the energies
>> will become stable. i.e.
>>
>> At line 558 in pb_force.f:
>>
>> change:
>>
>> if ( abs(totcrg) > smallcrg ) then
>>
>> to:
>>
>> if ( .false. and. abs(totcrg) > smallcrg ) then
>>
>> A patch will be posted at "AMBER Bug fixs" today.
>>
>> Thanks a lot for letting us know.
>>
>> Ray
>>
>> JunJun Liu wrote:
>>
>>> Dear Dr.Luo:
>>>
>>> Thank you for response! I find STD value decreased to about
>>> 5kcal/mol when using DelPhi program. I'd like to find out what the
>>> problem is, but I have to leave for about one week. I hope I still
>>> can get help from you one week later when I come back to this issue.
>>>
>>> Best Regards!
>>>
>>> Liu
>>>
>>> On Sun, 14 Aug 2005 03:40:54 +0800, Ray Luo <rluo_at_uci.edu> wrote:
>>>
>>>> Dear Liu,
>>>>
>>>> The origin in fdpb is defined as the center of a rectangular box
>>>> that is large enough for all atoms currently considered. You can
>>>> print out the center of grid used for each snapshot by setting
>>>> "pbverb=1" in the pb portion of your "pbsa.in" file or insert it
>>>> in the pb section of your perl script. Do you see any warnings in
>>>> your output files?
>>>>
>>>> I have turned off the option for end users to define the center of
>>>> grid in Amber8, because we have not seen relative standard
>>>> deviations as large as yours after implementing the new method to
>>>> compute reaction field (solvation) energy using surface charges.
>>>> This used to be an issue for pb calculations using only one
>>>> conformation, not an issue for mmpbsa type of calculation that
>>>> uses many conformations.
>>>>
>>>> All the best,
>>>> Ray
>>>>
>>>> JunJun Liu wrote:
>>>>
>>>>> Dear AMBER Users:
>>>>>
>>>>> I obtained 75 snapshots from 750ps to 1500ps every 10ps. The
>>>>> MM-PBSA results are shown below.
>>>>> # DELTA
>>>>> # -----------------------
>>>>> # MEAN STD
>>>>> # =======================
>>>>> ELE -37.81 5.48
>>>>> VDW -70.03 3.67
>>>>> INT -0.00 0.00
>>>>> GAS -107.85 5.00
>>>>> PBSUR -6.54 0.27
>>>>> PBCAL 76.48 19.28
>>>>> PBSOL 69.94 19.30
>>>>> PBELE 38.67 18.51
>>>>> PBTOT -37.91 18.33
>>>>>
>>>>> As can be seen, PBCAL is quite inaccurate when compared STD value
>>>>> of GAS energy. I got a suggestion that the origin of grids need
>>>>> to be checked. If the origin was defined at the center of mass
>>>>> or center of charge, then the origins in the complex and free
>>>>> protein are not necessarily the same, which will cause
>>>>> inappropriate solvation calculations.
>>>>>
>>>>> If the origin was always defined at the position of the same atom
>>>>> of the protein, which means the origins in both the complex and
>>>>> free protein should be the same, then what other factors that
>>>>> could cause the inappropriately numerical PB solvation? Any
>>>>> suggestions will be highly appriecated!
>>>>>
>>>>> Best Regards!
>>>>>
>>>>> Liu
>>>>>
>>>>
>>>
>>>
>>>
>>
>
>
>

-- 
====================================================
Ray Luo, Ph.D.
Department of Molecular Biology and Biochemistry
University of California, Irvine, CA 92697-3900
Office: (949)824-9528         Lab: (949)824-9562
Fax: (949)824-8551          e-mail: rluo_at_uci.edu
Home page: http://rayl0.bio.uci.edu/rayl/
====================================================

----------------------------------------------------------------------- The AMBER Mail Reflector To post, send mail to amber_at_scripps.edu To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu