[gmx-users] R: Re: adding a new residue in the ff

[Justin A. Lemkul]

Anna Marabotti wrote:
> Dear Justin and Tsjerk,
> thank you very much for your last encouraging words...;-) I don't remember a
> time in my life where something that I needed, which was very complex to do,
> was already available...(apart from Gromacs package, of course!)
> About parameterization: sure I was intended to use the parameters for Cys
> residue as much as possible. I was thinking of adding all angles, bonds,
> dihedrals etc until the CB, and using the topology created by PRODRG only
> for SO3 moiety. If I understand well, Tsjerk suggests not to include the
> CA-CB-SG angle. Concerning charges, I was planning to calculate them using
> Antechamber, following Justin's suggestions about the problems of Prodrg.

Results from Antechamber directly are probably not sufficient as an end state 
for parameterization.  See our recent paper about this: 
http://pubs.acs.org/doi/abs/10.1021/ci100335w.  Antechamber gives you a 
reasonable starting point, but the calculations do not even replicate the most 
trivial species in the Gromos96 43A1 parameter set (nor would you necessarily 
expect them to).

> IIRC, apart from charges, bonds, angles, dihedrals etc. obtained with Prodrg
> were substantially correct, so I was thinking of adding directly these
> values to my ffbonded.itp. The atom types of SO3 seem to be already present

If they're "correct," then what do you need to add?  If they are already part of 
the parent force field (i.e., PRODRG chose correct values) then you do not need 
to re-define existing parameters (and in fact, grompp may raise a fatal error if 
you do).

> in the ffG43a1, since they are identified as SDSMO and OM. Do you think that
> with these data I would be able to add the bonds, angles etc to my
> ffbonded.itp, to aminoacids.rtp and all other files I have to modify in the
> ffG43a1 local forcefield? I don't think to be able to obtain an "exact"
> parameterization, but I hope it would be at least acceptable...

Your desired level of accuracy is up to you, but most careful reviewers do not 
consider partial efforts to be sufficient.  Close enough is not necessarily good 
enough.  The approximations or substitutions you make need to be justifiable, 
which I guess is about as good as anyone can recommend.

-Justin

> Many thanks 
> Anna
> 
> 
> ----------------------------------------------------------------------
> 
> Message: 1
> Date: Thu, 31 Mar 2011 19:27:17 +0200
> From: Tsjerk Wassenaar <tsjerkw at gmail.com>
> Subject: Re: [gmx-users] adding a new residue in the ff
> To: jalemkul at vt.edu, Discussion list for GROMACS users
> 	<gmx-users at gromacs.org>
> Message-ID:
> 	<AANLkTikae2L6DdpFsZjJXXLqEkLxWGd1HETWSkeMvv53 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
> 
> Hi,
> 
>> It would probably be easier, faster, and more accurate to just use most of
>> the parameters for Cys rather than try to have PRODRG re-create a
>> (potentially flawed) model of your compound.  The only new parameters are
>> related to SO3, so the rest should be identical to the Cys residue.
> 
> That's not a good idea. It'll only do for the backbone. The charge of
> CB will be quite different, as well as the CA-CB-SG angle, and for the
> rest there's nothing similar even.
> 
>> This is where parameterization becomes a chore - when there's nothing
>> analogous to what you're doing.  Proper Gromos96 parameterization
>> methodology would dictate that you generate an analogous compound (i.e.,
>> methyl sulfonic acid) and adjust its parameters such that it reproduces
>> various condensed-phase thermodynamic and physical properties (DeltaG of
>> solvation, liquid density, heat of vaporization).  Proper derivation is
>> quite time-consuming.  Perhaps someone has already done this work and it
> is
>> published.  If you're unlucky, you've got to do it all yourself.
> 
> That's a good idea :)
> 
> Cheers,
> 
> Tsjerk
> 
> 
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================