[gmx-users] Discrepancy between Free Energy Perturbation using Gromos96 and OPLS

Erik Lindahl lindahl at sbc.su.se
Wed Oct 12 22:18:22 CEST 2005 

Hi,

A bug indeed. Fixed in both branches of CVS, and will be in a 3.3.1  
update in your near future.

In the meantime, you can download an updated src/kernel/toppush.c from

ftp://ftp.gromacs.org/pub/tmp/toppush.c


Cheers,

Erik


On Oct 12, 2005, at 11:21 AM, Erik Lindahl wrote:

> Hi Sandeep,
>
> This could indeed be a bug. Please enter a report at http:// 
> bugzilla.gromacs.org and include the appropriate files to reproduce  
> it as attachments.
>
> We'll probably have a look at it anyway, but it's going to be  
> longer if we need to construct a system to reproduce it first :-)
>
> Cheers,
>
> Erik
>
> On Oct 12, 2005, at 6:01 AM, Sandeep Somani wrote:
>
>
>> Hi All
>>
>> I have a clue on the origin of this discrepancy. Looks like a bug  
>> in FEP
>> code.
>> Would like to get comments from the developers.
>>
>> The system I am working with is a 153 residue protein with a buried
>> (isolated from solvent) water molecule. I am trying to find the free
>> energy of binding of this buried water by perturbing the non-bonded
>> interaction of the *single water molecule*.
>>
>> The deltaG values using different ff and gmx3.3beta are
>>
>> 1. OPLS ff for protein    :  ~ -330 kJ/mol
>> 2. Gromos96 (ffG43a1.itp) :  ~  -37 kJ/mol
>>
>> Clearly OPLS is incorrect.
>>
>> Using -sepdgdl option of mdrun in gmx3.3beta to get the components of
>> dG/dl,
>> a sample entry in the log file at lambda=0.5 looks like
>>
>> GROMOS
>> ======
>> Step 1000: non-bonded V and dVdl for node 0:
>>   VdW and Coulomb SR particle-p. V -3.34460e+05  dVdl  8.61080e+01
>>   RF exclusion correction        V -1.01159e+04  dVdl  5.09968e-02
>> Step 1000: bonded V and dVdl for node 0:
>>   G96Angle                #2306  V  2.50156e+03  dVdl  0.00000e+00
>>   Proper Dih.             # 857  V  1.30637e+03  dVdl  0.00000e+00
>>   Improper Dih.           # 743  V  7.88975e+02  dVdl  0.00000e+00
>>   LJ-14 + Coulomb-14      #2561                  dVdl  0.00000e+00
>> ------
>> OPLS
>> ====
>> Step 1000: non-bonded V and dVdl for node 0:
>>   VdW and Coulomb SR particle-p. V -3.42520e+05  dVdl  1.15979e+02
>>   RF exclusion correction        V -1.55339e+04  dVdl  7.03910e-02
>> Step 1000: bonded V and dVdl for node 0:
>>   Angle                   #4465  V  4.88804e+03  dVdl  0.00000e+00
>>   Proper Dih.             # 467  V  2.79401e+02  dVdl  0.00000e+00
>>   Ryckaert-Bell.          #5037  V  1.80946e+03  dVdl *-3.39702e+02*
>>   LJ-14 + Coulomb-14      #6445                  dVdl  0.00000e+00
>> --------
>>
>> The numbers are reasonable for Gromos.
>> But in OPLS there is a big and non-zero "Ryckaert-Bell." term even
>> though there are no dihedrals
>> in the perturbed atoms. Whats the origin of this ? And how can i  
>> get rid
>> of it ?
>> Relevent portions of the OPLS topology file are attached.
>>
>> also, in the log file what do the numbers starting with # mean  
>> (#4465, #
>> 467 etc) ?
>>
>> With Regards
>> Sandeep
>>
>> ---
>> Research Associate
>> Computational Biology Group
>> Bioinformatics Institute
>> Singapore
>>
>> http://www.bii.a-star.edu.sg/~ssomani
>>
>>
>>
>>
>>
>>
>>
>>
>>> -----Original Message-----
>>> From: gmx-users-bounces at gromacs.org
>>> [mailto:gmx-users-bounces at gromacs.org] On Behalf Of Sandeep Somani
>>> Sent: Tuesday, October 04, 2005 5:57 PM
>>> To: 'Discussion list for GROMACS users'
>>> Subject: RE: [gmx-users] Discrepancy between Free Energy
>>> PerturbationusingGromos96 and OPLS
>>>
>>>
>>> Adding to that, as per David's suggestion, I installed
>>> gmx3.3_rc3 and ran the same simulation again.
>>>
>>> The values are still similar i.e. ~ -30kJ/mol for Gromos96
>>> and < -300kJ/mol for OPLS.
>>>
>>> I tried again with PME too .. still the OPLS values are much
>>> lower than Gromos.
>>>
>>> am quite clueless now ! ill be happy to provide the input
>>> files if anyone wants to probe further ..
>>>
>>> Regards
>>> Sandeep
>>>
>>>
>>>
>>>
>>>
>>>
>>>> -----Original Message-----
>>>> From: gmx-users-bounces at gromacs.org
>>>> [mailto:gmx-users-bounces at gromacs.org] On Behalf Of Maik Goette
>>>> Sent: Tuesday, October 04, 2005 5:41 PM
>>>> To: Discussion list for GROMACS users
>>>> Subject: Re: [gmx-users] Discrepancy between Free Energy
>>>> Perturbation usingGromos96 and OPLS
>>>>
>>>>
>>>> Hi David
>>>>
>>>> Could you be a bit more precise?
>>>> Is this difference due to a problem of united
>>>>
>>>>
>>> atoms/reaction field or
>>>
>>>
>>>> all atoms/reaction field. For I also use FEP with all atoms
>>>> (amber99-port) and reaction field, I could use some info. :)
>>>>
>>>> Regards
>>>>
>>>> Maik Goette, Dipl. Biol.
>>>> Max Planck Institute for Biophysical Chemistry
>>>> Theoretical & computational biophysics department
>>>> Am Fassberg 11
>>>> 37077 Goettingen
>>>> Germany
>>>> Tel.  : ++49 551 201 2310
>>>> Fax   : ++49 551 201 2302
>>>> Email : mgoette[at]mpi-bpc.mpg.de
>>>>          mgoette2[at]gwdg.de
>>>> WWW   : http://www.mpibpc.gwdg.de/groups/grubmueller/
>>>>
>>>>
>>>> David wrote:
>>>>
>>>>
>>>>> On Sat, 2005-10-01 at 13:11 +0800, Sandeep Somani wrote:
>>>>>
>>>>>
>>>>>
>>>>>> hi all
>>>>>>
>>>>>> I did a Slow Growth (dummy -> real) based FEP of a small
>>>>>>
>>>>>>
>>>> ligand bound
>>>>
>>>>
>>>>>> to a protein using Gromos and OPLS ff. Description of
>>>>>>
>>>>>>
>>>> ligand in both
>>>>
>>>>
>>>>>> cases is identical.
>>>>>>
>>>>>> Free energy change from the Gromos simulation is
>>>>>>
>>>>>>
>>> reasonable but the
>>>
>>>
>>>>>> number from OPLS is an order of magnitude more negative.
>>>>>>
>>>>>> All parameters (reaction field electrostatics, constraints in
>>>>>> topology, delta-lambda etc) other than the ff are
>>>>>>
>>>>>>
>>>> identical. Am using
>>>>
>>>>
>>>>>> gmx3.2.1.
>>>>>>
>>>>>> Has anyone encountered this before ? what could be the
>>>>>>
>>>>>>
>>>> origin of this
>>>>
>>>>
>>>>>> discrepancy ?
>>>>>>
>>>>>>
>>>>>
>>>>> It could be due to the all-atom/united atom difference in
>>>>>
>>>>>
>>>> combination
>>>>
>>>>
>>>>> with the reaction field. Berk has recently implemented an
>>>>>
>>>>>
>>> exclusion
>>>
>>>
>>>>> correction for reaction fields that could be important
>>>>>
>>>>>
>>>> here. You could
>>>>
>>>>
>>>>> retry it with 3.3beta to see whether there is a difference.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>> Thanks and Regards
>>>>>> Sandeep
>>>>>>
>>>>>> ---
>>>>>> Research Associate
>>>>>> Computational Biology Group
>>>>>> Bioinformatics Institute
>>>>>> Singapore
>>>>>>
>>>>>> http://web.bii.a-star.edu.sg/~ssomani/
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> gmx-users mailing list
>>>>>> gmx-users at gromacs.org
>>>>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>>>>>> Please don't post (un)subscribe requests to the list. Use the
>>>>>> www interface or send it to gmx-users-request at gromacs.org.
>>>>>>
>>>>>>
>>>> _______________________________________________
>>>> gmx-users mailing list
>>>> gmx-users at gromacs.org
>>>> http://www.gromacs.org/mailman/listinfo/gmx-> users
>>>> Please
>>>> don't post (un)subscribe requests to the list. Use the
>>>> www interface or send it to gmx-users-request at gromacs.org.
>>>>
>>>>
>>>>
>>>
>>>
>>> _______________________________________________
>>> gmx-users mailing list
>>> gmx-users at gromacs.org
>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>>> Please don't post (un)subscribe requests to the list. Use the
>>> www interface or send it to gmx-users-request at gromacs.org.
>>>
>>>
>>> <opls_prot.top>
>>>
>> _______________________________________________
>> gmx-users mailing list
>> gmx-users at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-users
>> Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-request at gromacs.org.
>>
>
> -----------------------------------------------------------
> Erik Lindahl  <lindahl at sbc.su.se>
> Assistant Professor, Stockholm Bioinformatics Center
> Stockholm University, SE 106 91 Stockholm
> Phone: +46 8 5537 8564     Fax: +46 8 5537 8214
>
>
>
>
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use thewww  
> interface or send it to gmx-users-request at gromacs.org.
>
>

-----------------------------------------------------------
Erik Lindahl  <lindahl at sbc.su.se>
Assistant Professor, Stockholm Bioinformatics Center
Stockholm University, SE 106 91 Stockholm
Phone: +46 8 5537 8564     Fax: +46 8 5537 8214

More information about the gromacs.org_gmx-users mailing list