Re: [AMBER] best Generalized Born solvation model for intrinsically disordered peptides

From: Carlos Simmerling <carlos.simmerling.gmail.com>
Date: Thu, 5 Sep 2013 12:01:49 -0400

Thomas,
here are some of my thoughts on this...
if you look at the igb8 article we show that indeed there does remain some
modest helical overstabilization relative to TIP3P simulations (when both
are highly converged). It would probably be most evident in systems with
free energies close to zero, such as yours. Of course we've done no
extensive testing with the protein force field you mention, but looking at
the 99SB results in the igb8 paper might give you some extra insight into
what to expect. 99SB is slightly underhelical in my experience, so works
well with igb8, with a bit of error cancellation. Both errors are probably
a couple tenths of kcals. If you use a force field that is more helical
than 99SB, then indeed you might find that igb8 gives too much helix
compared to experiment. The GB models we use do have limitations, and it's
important to know what those are when interpreting results, especially for
systems with very weak intrinsic structure preferences where any small bias
can easily shift the equilibrium. With more stable structures, a few tenths
of kcal shift tends to have less apparent impact on the equilibrium
structure ratios (because of the log scale).

 What's important here I think is that usually there is a tradeoff between
accuracy and ease of sampling (and GPU performance these days). The GB
models will cost you some accuracy but usually give you much improved
precision. If you've got ways to sample in explicit solvent that give you
the precision that you need, then almost certainly you're better off doing
that than using a more approximate model. In my opinion, one of the keys to
success in this field is still in making judgments about when certain
approximations are reasonable for the specific scientific goals of a given
project. Even a "good" result is usually the result of some well-designed
compromises....
hope that helps.
carlos



On Thu, Sep 5, 2013 at 11:36 AM, Thomas Evangelidis <tevang3.gmail.com>wrote:

> Hi,
>
> I have run 4 explicit solvent aMD simulation (99SB-NMR1-ILDN & TIP4P-Ew) of
> an IDP with 2 disordered tails, starting from extended tail conformations.
> Each one was for 250 ns. In all conformations the first tail remained
> disordered with some helical turns, whereas the second adopted a helical
> conformation near the end of the simulation. In contrast during aMD
> simulations with igb 8 both disordered tails formed perfect helices and
> remained like that no matter how long the simulation was. Hence I'm pretty
> sure that, at least in my cases, igb 8 overestimates the helical content of
> the ID proteins/peptides.
>
> ~Thomas
>
> PS: I never said I ran just 17 ns of cMD in implicit solvent. What I said
> was that igb 8 folds the peptide to an alpha-helix after ~17 ns. The total
> simulation time was much longer.
>
>
> On 5 September 2013 17:48, Daniel Roe <daniel.r.roe.gmail.com> wrote:
>
> > Hi,
> >
> > On Tue, Sep 3, 2013 at 2:55 PM, Hai Nguyen <nhai.qn.gmail.com> wrote:
> > > how long you have been running? 20ns? if yes, I suggest to run much
> > longer
> > > like microsecond. For a small peptide, getting this long simulation
> isn't
> >
> > Just to expand on Hai's point a bit, one of the major issues when
> > comparing implicit solvent to explicit solvent runs is that in
> > explicit solvent, conformational sampling is typically much slower
> > than in implicit solvent (due to friction, explicit solvent
> > rearrangement, etc). You haven't said how long the explicit solvent
> > runs are, but you need to be sure that the structural properties in
> > each simulation are reasonably converged before you compare them, and
> > if you've only run explicit solvent as much as your implicit (17 ns)
> > chances are the run is not converged.
> >
> > > the parameters quoted at the end of the message. In contrast, when
> using
> > > igb 7 the peptide has some helical propensity but is mostly
> > > unstructured (helical and random coil conformations are in
> equilibrium).
> >
> > This result may not necessarily be wrong, but as Carlos said the GB
> > neck (igb=7) model has a tendency to destabilize hydrogen bonding so
> > be wary of these results. For more on this see:
> >
> > Roe, D.R.; Okur, A.; Wickstrom, L.; Hornak, V.; Simmerling, C., J.
> > Phys. Chem. B, 2007, 111, 18461857.
> > Shell, M.S.; Ritterson, R.; Dill, K.A., J. Phys. Chem. B., 2008, 112,
> > 6878-6886.
> >
> > -Dan
> >
> > --
> > -------------------------
> > Daniel R. Roe, PhD
> > Department of Medicinal Chemistry
> > University of Utah
> > 30 South 2000 East, Room 201
> > Salt Lake City, UT 84112-5820
> > http://home.chpc.utah.edu/~cheatham/
> > (801) 587-9652
> > (801) 585-9119 (Fax)
> >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
>
>
> --
>
> ======================================================================
>
> Thomas Evangelidis
>
> PhD student
> University of Athens
> Faculty of Pharmacy
> Department of Pharmaceutical Chemistry
> Panepistimioupoli-Zografou
> 157 71 Athens
> GREECE
>
> email: tevang.pharm.uoa.gr
>
> tevang3.gmail.com
>
>
> website: https://sites.google.com/site/thomasevangelidishomepage/
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Sep 05 2013 - 09:30:02 PDT
Custom Search