Dear Shuntaro,
> I have tried the both approaches and found differences.
> I used CH3SOCH3 as an example target molecule.
> RESP charges of the resultant mol2 files are as follows:
>
> epsilon=4.0 approach > Solvent=Ether approach
> 1 C1x y z C1 -0.1752 **** versus > 1 C1x y z C1 -0.1749 ****
> 2 H2x y z H2 0.1040 **** versus > 2 H2x y z H2 0.1042 ****
> 3 H3x y z H3 0.1040 **** versus > 3 H3x y z H3 0.1042 ****
> 4 H4x y z H4 0.1040 **** versus > 4 H4x y z H4 0.1042 ****
> 5 S5x y z S5 0.2220 **** versus > 5 S5x y z S5 0.2216 ****
> 6 O6x y z O6 -0.4844 **** versus > 6 O6x y z O6 -0.4860 ****
> 7 C7x y z C7 -0.2301 **** versus > 7 C7x y z C7 -0.2294 ****
> 8 H8x y z H8 0.1186 **** versus > 8 H8x y z H8 0.1187 ****
> 9 H9x y z H9 0.1186 **** versus > 9 H9x y z H9 0.1187 ****
> 10 H10x y z H10 0.1186 **** versus > 10 H10x y z H10 0.1187 ****
>
> where, I omitted actual coordinates.
> If a target molecule is charged, differences may be larger.
> I have not calculated that kind of molecules, though.
Two remarks:
- The two methyl groups do not bear the same charge values. I guess
you used Ante_R.E.D. 1.4, which does know how to deal with chemical
equivalencing between different chemical groups. Use R.E.D.
Server/Ante_R.E.D. 2.0 to solve this problem. Please, read:
http://q4md-forcefieldtools.org/REDS/popup/popanteredtopequiv.php
- The differences you report seems quite small...
> Although I have no idea which way amber 03 was actually derived,
> I have found a document where Yong Duan referred to the subject.
> http://structbio.vanderbilt.edu/archives/amber-archive/2005/1290.php
> It seems Solvent=Ether would be appropriate.
'appropriate'...: at least, it seems this is the g03 keyword that was used ;-)
If you continue to read this '1290.php' email, you can see that ff03
charge values can not be reproduced. This point has been discussed
many times in the Amber mailing list, and this is the main reason why
R.E.D. (and then R.E.DD.B. & R.E.D. Server) was develop: to derive
reproducible charges; your charge values will be reproducible by other
researchers only if you use R.E.D.
regards, Francois
> (2011/05/18 20:17), FyD wrote:
>> Shuntaro,
>>
>>>> Why not using R.E.D. Server? any researcher can use 8 cpu cores...
>>>>
>>> Thank you for your offer.
>>> However my target molecule consists of over 100 atoms
>>> and number of initial structures will be no less than 10.
>>
>> - R.E.D. Server . http://q4md-forcefieldtools.org/REDS/ is limited to
>> 150 atoms
>> - R.E.D. Server Development @
>> http://q4md-forcefieldtools.org/REDS-development/ is limited to 350
>> atoms; but... You should consider spilling your big molecule into more
>> elementary(smaller) building blocks.
>>
>>> By the way, the following lines are quoted from the Gaussian input
>>> that RED-vIII.4.pl (to which Corn2Duan.csh have been already applied)
>>> generates.
>>
>> ok
>>
>>> ‘#P b3lyp/cc-pVTZ SCRF(IEFPCM,Solvent=Ether) SCF(Conver=6) NoSymm Test
>>> Pop=mk IOp(6/33=2) GFInput GFPrint’
>>
>> If I remember the SCRF'IEFPCM' solvation model in G03 is different to
>> that in g09: I would use in your case, Gaussian 2003.
>>
>>> Why do they use Solvent=Ether,
>>> not SCRF(IEFPCM, Read) with input section for PCM keywords (eps=4.0)?
>>> In deriving RESP charges compatible with Amber 03 force field,
>>> we should use B3LYP/cc-pVTZ method with IEFPCM (epsilon=4.0).
>>> (Duan et al. Vol. 24, No. 16. Journal of Computational Chemistry)
>>
>> I see what you mean...
>>
>> . http://www.gaussian.com/g_tech/g_ur/k_scrf.htm
>> DiethylEther: ε=4.2400
>> . http://en.wikipedia.org/wiki/Diethyl_ether_(data_page)
>> DiethylEther: ε0 = 4.34 at 20 °C
>>
>> Did you try to use both approaches? (I am sure we tested that, but I
>> forgot...)
>> Is there any difference?
>>
>> regards, Francois
>>
>>
>>> (2011/05/18 18:03), FyD wrote:
>>>> Dear Suntaro,
>>>>
>>>>> Thank you for your reply.
>>>>> I have introduced R.E.D. III.4 tools with g03 rev E.1.
>>>>>
>>>>> I want to derive RESP charges of a new residue which are compatible
>>>>> with
>>>>> Amber 03 force field using many initial conformations.
>>>>> So I am conducting the series of commands described in
>>>>> ‘http://q4md-forcefieldtools.org/Tutorial/Tutorial-1.php’.
>>>>> However, I bumped into a problem in the stage of Gaussian jobs of
>>>>> ‘perl RED-vIII.4.pl’. The problem is that it takes too much time.
>>>> Why not using R.E.D. Server? any researcher can use 8 cpu cores...
>>>>
>>>>> So I am going to execute the Gaussian jobs in advance
>>>>> where the log files are named after those RED-vII.4.pl will make,
>>>>> then execute RED-vIII.4.pl with commenting out
>>>>> “system("$gauss JOB1-gau_m$NM-$NC.com");” in RED-vIII.4.pl.
>>>>> Can this have a bad effect on the procedure for making RESP charges?
>>>> Yes, you can do that (however, provide the correct file name_s_ IN
>>>> AGREEMENT WITH THE P2N FILE)- we did exactly that for a user ;-)
>>>> This corresponds to a pseudo "Re-fit" mode...
>>>>
>>>> regards, Francois
>>>>
>>>>
>>>>> (2011/05/16 15:07), FyD wrote:
>>>>>> Dear Chiba Suntaro,
>>>>>>
>>>>>>> I am deriving RESP charges of a new residue using Gaussian 09 rev
>>>>>>> B.01
>>>>>>> (g09).
>>>>>>>
>>>>>>> According to ‘AmberTools Bug Fixes
>>>>>>> (http://ambermd.org/bugfixesat.html)
>>>>>>> <http://ambermd.org/bugfixesat.html%29>‘,
>>>>>>> “In Gaussian09 rev B.01, the facility to write out the electrostatic
>>>>>>> potential
>>>>>>> on a grid of points was inadvertently deleted.”
>>>>>> - You could use the R.E.D. III.4 tools with GAMESS or Firefly, or
>>>>>> - You could also use R.E.D. Server 2.0, which interfaces GAMESS,
>>>>>> Firefly or Gaussian 09 version A2 (or Gaussian 3003 version E.01).
>>>>>>
>>>>>> See http://q4md-forcefieldtools.org/
>>>>>>
>>>>>> regards, Francois
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 20 2011 - 01:00:03 PDT
