Hi Romelia,
Thank you for your suggestions. I believe they will save other people as
well from "disoriented" experimentation with the aMD parameters.
Each monomer has 80 residues and I 'm using the Amber99SB-NMR1-ILDN ff. One
important detail I omitted is that my test simulations were done in
implicit solvent because I get 7x fold speed up compared with the original
system, which has TIP4P-Ew waters and ions and counts 134000 atoms in
total. Today I ran a couple of ns in explicit solvent and noticed that the
dimer is more stable, actually it doesn't move that much with the suggested
parameter values. I will carry out more tests in explicit solvent and let
you know about the outcome.
Just one last question, when I estimate Ep and aP values should I count
each TIP4P-Ew water as 4 atoms or as 3? So far I counted each water as 4
atoms.
Thank you so much!
Thomas
On 19 November 2012 20:20, Romelia Salomon <romelia.caltech.edu> wrote:
> Hi Thomas
>
> Try using these parameters:
> Ed=Ed_average+3.5*ResNum (smaller per residue boost which will be a
> smaller Ed and less agressive)
> aD=3.5*ResNum*0.25 (larger scale factor which will be a larger alpha
> which will be LESS agressive)
>
> Ep=Ep_average+0.16*AtomNum (this is probably okay)
> aP=0.16*AtomNum (this is okay)
>
>
> What FF is you are using and how many residues is one of the monomers? In
> a collaboration we have some runs with HIVPR (which is a dimer) with
> ff99SB and using the above suggested parameters and we have not seen a
> separation of the dimer. You should see more movement in your structure
> and that the secondary structures are change a bit, but they should not
> unfold in an unexpected way (I am working on a protein that undergoes a
> large transition and part of it involves the unfolding of an alpha helix).
> Is it an explicit solvent or GB?
>
> Maybe it is in the way the system is being initialized.
>
>
> Best,
> Romelia
>
> > Hi Romelia,
> >
> > It happens both, but the dissociation of the dimer is more noticeable.
> The
> > dimer is a helical bundle and as the helices start to bend it
> dissociates.
> >
> > thanks,
> > Thomas
> >
> >
> > On 19 November 2012 18:35, Romelia Salomon <romelia.caltech.edu> wrote:
> >
> >> Dear Thomas
> >>
> >> One quick question first, is it unfolding the secondary structures or
> >> separating the two components of the dimer?
> >>
> >> Best wishes,
> >>
> >> Romelia
> >>
> >> > Dear Romelia and other AMBER users/developers,
> >> >
> >> > Thanks for the clarification. I have tried the rule of thump to select
> >> E
> >> > and alpha values, which is described both in the recent paper you
> >> cited
> >> > and
> >> > the manual. Namely, to decide the dihedral boost parameters (Ed, aD)
> >> and
> >> > the potential boost parameters (Ep, aP) I did:
> >> >
> >> > Ed=Ed_average+4*ResNum
> >> > aD=4*ResNum*0.2
> >> > Ep=Ep_average+0.16*AtomNum
> >> > aP=0.16*AtomNum
> >> >
> >> > However, these parameter values lead to the unfolding of my protein
> >> dimer.
> >> > I have also tried setting aD=4*ResNum*0.4 but I still observe the same
> >> > effect.
> >> > How would you recommend making the boost smoother? By increasing just
> >> the
> >> > alpha values? Do I need to increase both of them?
> >> >
> >> > thanks in advance,
> >> > Thomas
> >> >
> >> >
> >> >
> >> >
> >> > On 27 September 2012 01:27, Romelia Salomon <romelia.caltech.edu>
> >> wrote:
> >> >
> >> >> Hi Thomas
> >> >>
> >> >> I added comments to your questions bellow.
> >> >>
> >> >> > Dear Amber users,
> >> >> >
> >> >> > I want to monitor the dynamics between a hetero-dimeric helical
> >> bundle
> >> >> and
> >> >> > a peptide. It is believed that the peptide binds at the
> >> >> protein-protein
> >> >> interface of the dimer and unwinds the helix of one of the
> >> components.
> >> >> For
> >> >> > this purpose I am considering accelerated MD (aMD) but have some
> >> >> queries
> >> >> about it:
> >> >> >
> >> >> > 1. Unlike Metadynamics, the bias in the potential is not history
> >> >> dependent,
> >> >> > therefore the system may explore multiple times the same area of
> >> >> conformational space. Is this correct?
> >> >>
> >> >> This is correct, AMD will make it more probable to move out of deep
> >> >> stable
> >> >> basins in the conformational space and there is no mechanism to
> >> prevent
> >> >> it
> >> >> from going back.
> >> >>
> >> >> >
> >> >> > 2. Is it possible to find the predominant conformations of the
> >> complex
> >> >> from
> >> >> > an aMD trajectory, namely the ones that we would expect to observe
> >> the
> >> >> most
> >> >> > in an unbiased MD trajectory?
> >> >>
> >> >> Yes, what you will get in the end of the AMD calculation is a state
> >> >> distribution, that although is not the unbiased (canonical) one, it
> >> >> still
> >> >> holds some resemblance to it. AMD distributions can be re-weighted to
> >> >> produce Boltzmann distributions.
> >> >>
> >> >> Please refer to these papers for more information.
> >> >> (a recent paper we just published on the GPU implementation of AMD)
> >> >> http://pubs.acs.org/doi/abs/10.1021/ct300284c
> >> >> (one of the main papers for this method)
> >> >> http://www.cse.nd.edu/~izaguirr/HaMM04.pdf
> >> >>
> >> >>
> >> >> I hope this helps, let me know if you need more information.
> >> >>
> >> >> Best,
> >> >>
> >> >> Romelia
> >> >
> >>
> >>
> >> --
> >> ****************************************
> >> Romelia Salomon
> >> Walker Group
> >> 398 San Diego Supercomputing Center
> >> UC San Diego
> >>
> >>
> >
> >
> > --
> >
> > ======================================================================
> >
> > 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/
> >
>
>
> --
> ****************************************
> Romelia Salomon
> Walker Group
> 398 San Diego Supercomputing Center
> UC San Diego
>
>
--
======================================================================
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
Received on Mon Nov 19 2012 - 12:00:02 PST
