AMBER Archive (2009)

Subject: Re: [AMBER] IGB=7 vs IGB=5

From: Carlos Simmerling (carlos.simmerling_at_gmail.com)
Date: Tue Jul 21 2009 - 08:51:22 CDT

sorry for the late reply, I originally missed this post.
in the JPC-B paper you refer to below, we showed that the igb=7 model had
very high cancellation of error, and also the radii errors demonstrate that
desolvation penalties in the current models are likely too high. in our
experience, the current igb=7 parameters tend to overly weaken secondary
structure/hydrogen bond formation as compared to explicit solvent
simulations. if you are going to use it, you should proceed carefully and
make sure to validate the results against experimental data.

On Thu, Jun 25, 2009 at 7:20 AM, <guardiani_at_fi.infn.it> wrote:

> Dear Amber experts,
>
> I am currently performing REMD simulations of a glycopeptide in
> implicit solvent. So far I have been using the IGB=7 Generalized
> Born model, but I am wondering whether it could be more appropriate
> to use IGB=5. In particular, answering to a different question to
> the mailing list, Ross Walker suggested using IGB=5 since
>
> "... the general consensus right now is that IGB=7 does not improve
> over the previous GB results and can potentially make things worse."
>
> I had chosen to use IGB=7 as I had read in
>
> J. Chem. Theory Comput. 2007, 3, 156-169 and in
> J. Phys. Chem. B 2007, 111, 1846-1857
>
> that this particular implementation of the GB model is not affected
> by a bias towards alpha-heical structures, as was the case of previous
> implementations such as the OBC GB model.
>
> I would like to know if there is any recent paper discussing the drawbacks
> and limitations of the IGB=7 implementation of the GB model, and I would
> also like to know why there is a general consensus on the fact that
> IGB=7 may potentially make things worse as compared to previous
> implementations
> of the GB model.
>
> I thank you very much for your help.
>
> Best regards,
>
> Carlo Guardiani
>
>
>
>
>
>
>
>
> Quoting Ross Walker <ross_at_rosswalker.co.uk>:
>
> Hi Carlo,
>>
>> Now my parametrization is as simple as that (I am using the bondi
>>> radii because I want to try an implicit-solvent simulation with
>>> igb=7, but of course I will also run the explicit-solvent one):
>>>
>>
>> Unless you have specific reasons for using IGB=7, say comparison to
>> previous
>> results, then I would suggest avoiding it and using IGB=5. I think the
>> general consensus right now is that IGB=7 does not improve over the
>> previous
>> GB results and can potentially make things worse.
>>
>> xleap -s -f $AMBERHOME/dat/leap/cmd/leaprc.ff99SB
>>>
>>> set default PBradii bondi
>>> source leaprc.glycam04
>>> x = sequence { NVAL GLU ARG NLN GLY HIS CSER }
>>> set x tail x.4.11
>>> set 1GB head 1GB.1.1
>>> y = sequence { x 1GB }
>>> savepdb y y.pdb
>>> saveamberparm y y.top y.crd
>>> quit
>>>
>>>
>>> Please let me know if this is what you meant. Which charges I will be
>>> using with this parametrization ? Charges based on the Connolly surface
>>> for the amino-acid residues and charges based on the CHELPG algorithm
>>> for the sugar ?
>>>
>>
>> I would use RESP charges for everything. See: W.D. Cornell, P. Cieplak,
>> C.I.
>> Bayly & P.A. Kollman J. Am. Chem. Soc. 1993, 115, 9620-9631.
>>
>> As for the 1-4 scaling factors, I performed my previous simulations
>>> using the default values, e.g.
>>>
>>> SCNB = 2.0
>>> SCEF = 1.2
>>>
>>> Are these values appropriate for a glycopeptide simulation ?
>>> If I explicitly set these values in the .in input files of
>>> my REMD simulation, will the apply to both the peptide and
>>> the sugar parts ?
>>>
>>
>> There is a slight mismatch right now between the Glycam force field and
>> the
>> protein ones. The protein force fields all scale VDW by 2.0 and EEL by 1.2
>> (I assume you meant SCEE above and not SCEF) which are the defaults.
>> GLYCAM
>> on the other hand does not do any scaling so would have SCNB = SCEE = 1.0.
>> AMBER 10 does not support such mixing of different 1-4 scaling factors
>> (AMBER 11 will support it) and thus at present you have to choose a
>> compromise. I believe the current 'accepted practice' is just to use the
>> defaults of 2.0 and 1.2 for everything when doing mixed protein /
>> carbohydrate simulations.
>>
>> Good luck,
>> Ross
>>
>> /\
>> \/
>> |\oss Walker
>>
>> | Assistant Research Professor |
>> | San Diego Supercomputer Center |
>> | Tel: +1 858 822 0854 | EMail:- ross_at_rosswalker.co.uk |
>> | http://www.rosswalker.co.uk | PGP Key available on request |
>>
>> Note: Electronic Mail is not secure, has no guarantee of delivery, may not
>> be read every day, and should not be used for urgent or sensitive issues.
>>
>>
>>
>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER_at_ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>>
>
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER_at_ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber