Re: Generalized Born solvation models supported by NAMD

From: Aron Broom (broomsday_at_gmail.com)
Date: Tue Aug 02 2016 - 09:41:09 CDT

Hi Chitrak,

I think there may be some dependence of the implicit solvent parameters on
the forcefield. I've always used AMBER forcefields, so I can't say for
sure.

I suspect there is some comparison of the models out there (in particular I
would imagine that when developing the newer models they compared to the
older ones), but I haven't seen an overwhelming consensus of one being
superior to the other.

Good luck!

Aron

On Fri, Jul 29, 2016 at 5:05 PM, Chitrak Gupta <chgupta_at_mix.wvu.edu> wrote:

> Hi Aron and Thomas,
>
> Thanks a lot for the discussion. I am getting started with implicit
> solvent simulations, so this helped a lot.
>
> One question I have is, when you switch between NAMD and AMBER, do you
> also switch the parameters you use (i.e. from CHARMM to AMBER)? In other
> words, does the implicit solvent have to be compatible with the force field
> that was used to parameterize the protein? Or could I use CHARMM parameters
> in AMBER with reliable results (with respect to GB model)?
>
> Also, between OBC models I and II, is either of them recommended to be
> "better" for peptide/water systems?
>
>
> Regards,
> Chitrak.
>
> On Fri, Jul 29, 2016 at 3:24 PM, Aron Broom <broomsday_at_gmail.com> wrote:
>
>> I have never done a comparison against the AMBER results, but I've had
>> good comparison to experiment when doing things with OpenMM.
>>
>> OpenMM runs extremely fast on GPUs with implicit solvent, comparable or
>> even slightly faster than AMBER in my experience, and certainly faster than
>> NAMD for these applications.
>>
>> Yes, OpenMM does support the HCT model. If you download the source (
>> https://simtk.org/projects/openmm), it comes with documentation, and you
>> can see the implicit solvent models available on page 16 of the
>> OpenMMUsersGuide.pdf.
>>
>> On Fri, Jul 29, 2016 at 2:18 PM, Thomas Evangelidis <tevang3_at_gmail.com>
>> wrote:
>>
>>>
>>> > No I haven't. Is that igb=1 in AMBER?
>>>
>>> Yes.
>>>
>>> > I did some implicit solvent stuff in OpenMM, which supports
>>> equivalents of igb=2,5 and 7. It runs quickly on GPUs.
>>>
>>> But do you know if it supports igb=1? I can rank the ligands of a
>>> specific protein better with this model, and I am looking for a commercial
>>> free code to run on GPUs. Have you noticed the perfomance of OpenMM to be
>>> better on GPUs than NAMD? Could you reproduce with OpenMM the results you
>>> obtained with AMBER using the afore-mentioned GB models?
>>>
>>> > I think the AMBER and OpenMM SASA calculations are a bit more
>>> approximate than those in NAMD, but are in a form that transfers well to
>>> the GPU.
>>>
>>> >
>>> > On Fri, Jul 29, 2016 at 12:20 PM, Thomas Evangelidis <
>>> tevang3_at_gmail.com> wrote:
>>> >>
>>> >>
>>> >>
>>> >> On 29 July 2016 at 17:55, Aron Broom <broomsday_at_gmail.com> wrote:
>>> >>>
>>> >>> I recall having this same question some years ago, and as I recall
>>> the answer was yes, you do need "sasa on".
>>> >>>
>>> >>> The method for calculating SASA that is used by NAMD (at least as of
>>> version 2.8) isn't exactly identical to that used by AMBER, so expect some
>>> difference in that energy term if you are comparing the two.
>>> >>>
>>> >>> I don't know if it is still the case, but as least in NAMD 2.8, the
>>> sasa calculation was very performance draining, at least when trying to
>>> accelerate the simulation using GPUs (I think the sasa calculations had not
>>> been ported to the GPU). Maybe that is no longer the case.
>>> >>>
>>> >>
>>> >> ​
>>> >> Thanks, they still slow down the performance a lot on GPUs!
>>> >> Have you ever tried to use the GB-HCT model in NAMD?
>>> >>
>>> >>
>>> >>
>>> >>>
>>> >>>
>>> >>> On Fri, Jul 29, 2016 at 11:26 AM, Thomas Evangelidis <
>>> tevang3_at_gmail.com> wrote:
>>> >>>>
>>> >>>> Greetings,
>>> >>>>
>>> >>>> I know NAMD supports the GB-OBC models I and II for implicit
>>> solvent simulations [1], which correspond to AMBER parameter igb=2 and
>>> igb=5 values, respectively. One simply has to change GBISDelta, GBISBeta,
>>> and GBISGamma values to switch between these two models. Is it necessary to
>>> activate surface tension (NAMD parameters "sasa on" and "surfaceTension
>>> 0.0072", as in AMBER)?
>>> >>>>
>>> >>>> Also, is it possible to use with NAMD the older
>>> >>>> ​​
>>> >>>> GB-HCT model [2], which corresponds to AMBER parameter igb=1?
>>> >>>>
>>> >>>> thanks in advance,
>>> >>>> Thomas
>>> >>>>
>>> >>>>
>>> >>>> ​[1] Onufriev, A.; Bashford, D.; Case, D. Exploring protein native
>>> states and large-scale conformational changes with a modified generalized
>>> Born model. Proteins, 2004, 55, 383–394.
>>> >>>> [2]​ Tsui, V.; Case, D. Molecular dynamics simulations of nucleic
>>> acids using a generalized Born solvation model. J. Am. Chem. Soc., 2000,
>>> 122, 2489–2498.
>>> >>>>
>>> >>>>
>>> >>
>>> >> --
>>> >>
>>> >> ======================================================================
>>> >>
>>> >> Thomas Evangelidis
>>> >>
>>> >> Research Specialist
>>> >>
>>> >> CEITEC - Central European Institute of Technology
>>> >> Masaryk University
>>> >> Kamenice 5/A35/1S081,
>>> >> 62500 Brno, Czech Republic
>>> >>
>>> >> email: tevang_at_pharm.uoa.gr
>>> >>
>>> >> tevang3_at_gmail.com
>>> >>
>>> >>
>>> >> website: https://sites.google.com/site/thomasevangelidishomepage/
>>> >>
>>> >>
>>> >
>>> >
>>> >
>>> > --
>>> > Aron Broom M.Sc
>>> > PhD Student
>>> > Department of Chemistry
>>> > University of Waterloo
>>>
>>
>>
>>
>> --
>> Aron Broom M.Sc
>> PhD Student
>> Department of Chemistry
>> University of Waterloo
>>
>
>

-- 
Aron Broom M.Sc
PhD Student
Department of Chemistry
University of Waterloo

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:36 CST