Re: [AMBER] Loop refirement

From: Carlos Simmerling <carlos.simmerling.gmail.com>
Date: Wed, 16 Jul 2014 11:37:04 -0400

I think that if you choose a reasonably accurate GB model then this task is
much easier than with explicit water. Explicit may be more accurate, but
sampling loop conformational changes can be too slow. You always have to
trade off accuracy and sampling. I would suggest giving our igb=8 model a
try (read the manual for suggestions on radii, etc). As always, you'll want
to make sure you have some data against which you can validate any
predictions that you make about structure. Regarding the REMD part, it's
another good reason to give GB a try. REMD in explicit water is expensive
(many replicas) and quite slow. Freezing part could be a problem- I'm not
sure if you can do that in all of the Amber MD codes (using the GB pmemd
code is probably best). You could choose positional restraints on the
non-loop region to keep it fixed. Having it be totaly frozen might not be
good anyway, since there may be some adjustment needed for different loop
options.

Another possibility is to use loop modeling that doesn't involve MD - such
as the analytical approaches. Then you might rescore the various models
with a good MM+GBSA approach.
good luck
CS


On Thu, Jul 10, 2014 at 5:44 AM, James Starlight <jmsstarlight.gmail.com>
wrote:

> some suggestions.
>
> some people gave me evidence that for my task (see a full set of loop
> confirmations and chose most probable) it will not good to use implicit
> solvent +amd because this will produce very unphysical thermodynamics isn't
> it?
>
> In fact I'm dealing with the membrane protein where membrane-embeded part
> should be fixed (I would not refine something here) and loops which are
> exposed to the solvent must be free to move. In this regards I've tried to
> applied gb model of IS with the frozen of not refined part of my protein.
> Will it be reasonable to use REMD with such implicit solvent model for the
> refinement? How It could be possible to really simplify REMD protocol for
> such loop prediction (e,g using small number of replicas or not).
> Some another suggestion (e.g brut force md with gb models)?
>
> James
>
>
>
> 2014-07-09 12:01 GMT+02:00 James Starlight <jmsstarlight.gmail.com>:
>
> > some updating of my issue
> >
> > I need to refine regions of my model consisted of water exposed 10-15
> > residues loops in which I'm not certain after its homology modeling. For
> > this task I'd like to
> > 1) Freeze all atoms of the protein consisted of the secondary structure
> > elements in which I'm not interest.
> > 2)Use some implicit solvent model for this simulation.
> > 3) Use some enhancing sampling technique to sample all possible
> > conformation of the loops at short timescale but keeping initial
> > thermodynamics of the system => predict possible folding in the loops
> > during the refinement.
> >
> > please suggest me possible GB implicit solvent model as well as enhanced
> > sampling engine (I'm chosing between replica exchange and accelerated md
> > with dihedral boost only). Any additional methods?
> >
> > I'll be very thankful to all,
> >
> >
> > James
> >
> >
> > 2014-06-14 22:21 GMT+02:00 James Starlight <jmsstarlight.gmail.com>:
> >
> > Also I'll be thankful if someone check my example SA script with applied
> >> multiple position restraints to some segment of my protein (here I'd
> like
> >> to freeze all atoms but not loop which I'd like to sample).
> >>
> >> SA with posres
> >> &cntrl
> >> imin=0,
> >> ntx=1,
> >> irest=0,
> >> ntc=2,
> >> ntf=2,
> >> tol=0.0000001,
> >> nstlim=50000,
> >> ntt=3,
> >> gamma_ln=1.0,
> >> ntr=1,
> >> ig=-1,
> >> ntpr=100,
> >> ntwr=10000,
> >> ntwx=100,
> >> dt=0.002,
> >> nmropt=1,
> >> ntb=0,
> >> ntp=0,
> >> cut=999.0,
> >> ioutfm=1,
> >> ntxo=2,
> >> igb=1,
> >> /
> >> &wt
> >> type='TEMP0',
> >> istep1=0,
> >> istep2=10000,
> >> value1=0.0,
> >> value2=103.0 /
> >> &wt
> >> type='TEMP0',
> >> istep1=10001,
> >> istep2=20000,
> >> value1=103.0,
> >> value2=203.0 /
> >> &wt
> >> type='TEMP0',
> >> istep1=20001,
> >> istep2=50000,
> >> value1=203.0,
> >> value2=303.0 /
> >> &wt type='END' /
> >> fixed
> >> 1000.0
> >> RES 1 67
> >> END
> >> fixed
> >> 1000.0
> >> RES 75 142
> >> END
> >> fixed
> >> 1000.0
> >> RES 169 241
> >> END
> >> fixed
> >> 1000.0
> >> RES 249 286
> >> END
> >> END
> >>
> >>
> >> Here I try to heat my system in 3 subsequent steps performing simulation
> >> using implicit solvent without PBC. Does it correct in general? I could
> not
> >> visualize my system in VMD using
> >> vmd -parm7 b2ar_Amber.prmtop -netcdf sa.nc
> >> what should I fix here?
> >>
> >>
> >> James
> >>
> >>
> >> 2014-06-13 23:50 GMT+04:00 James Starlight <jmsstarlight.gmail.com>:
> >>
> >> Dear Vlad,
> >>>
> >>>
> >>> many thanks for suggestions. I've already seen some papers describing
> >>> some methodologies of structural refinement based of some enhanced
> sampling
> >>> methods. However in case of loop refinement what could be expected
> from the
> >>> brute-force md with aplied restraints on the rest of the protein
> (excluding
> >>> refined loops) using 1) implicit solvent 2) some
> high-temperatutre-based
> >>> method like simulating annealing.
> >>>
> >>> James
> >>>
> >>>
> >>> 2014-05-28 11:53 GMT+04:00 Vlad Cojocaru <
> >>> vlad.cojocaru.mpi-muenster.mpg.de>:
> >>>
> >>> Dear James,
> >>>>
> >>>> I am afraid you'd have to do some reading ... Its very hard to believe
> >>>> that somebody on this list has the time to give you detailed
> >>>> instructions. What you ask for is a summary of many different papers.
> >>>> The Amber manual has an example of simulated annealing protocol for
> NMR
> >>>> refinement which used to be with distance dependent dielectric (maybe
> it
> >>>> has changed in the meantime). Anyhow, you'd have to adapt that to the
> >>>> implicit solvent model you wish to use. The implicit solvent models
> are
> >>>> all well documented in the corresponding publications which are
> >>>> referenced in the Amber manual.
> >>>>
> >>>> Besides, take care how you interpret your results. The longer the
> loops,
> >>>> the less you can rely on the loop refinement. You'd need to run a
> number
> >>>> of different simulations, maybe even test different force fields ...
> >>>> Especially if loops are functionally important, you may easily draw
> >>>> wrong conclusions from such refinements. Comparison with experiments
> is
> >>>> always good.
> >>>>
> >>>> Best,
> >>>> Vlad
> >>>>
> >>>>
> >>>> On 05/28/2014 09:29 AM, James Starlight wrote:
> >>>> > I try to specify my question.
> >>>> >
> >>>> > I suppose that force field based simulated annealing with positions
> >>>> > restraints applied to the all protein atoms but not for loops which
> >>>> I'd
> >>>> > like to refine might be exactly what I'm looking for. Could someone
> >>>> suggest
> >>>> > appropriate SA setups for such loop refirement: e.g I'm interesting
> in
> >>>> > number of SA windows, coupling constants in each windows,
> appropriate
> >>>> > implicit solvent models?
> >>>> >
> >>>> >
> >>>> > James
> >>>> >
> >>>> >
> >>>> > 2014-05-26 14:06 GMT+04:00 James Starlight <jmsstarlight.gmail.com
> >:
> >>>> >
> >>>> >> Dear Amber's users!
> >>>> >>
> >>>> >>
> >>>> >> I need to refine some flexible regions (mainly long loop and linker
> >>>> >> regions) of my proteins prior to the production MD run using some
> >>>> enhanced
> >>>> >> sampling engines implemented in Amber like accelerated molecular
> >>>> dynamics
> >>>> >> or simulated annealing. Please provide me with some basic ideas of
> >>>> the
> >>>> >> easiliest realization of these methods in amber including suitable
> >>>> implicit
> >>>> >> solvent models for such task with the tutorials and further
> reading.
> >>>> >>
> >>>> >>
> >>>> >> TFH,
> >>>> >>
> >>>> >> James
> >>>> >>
> >>>> > _______________________________________________
> >>>> > AMBER mailing list
> >>>> > AMBER.ambermd.org
> >>>> > http://lists.ambermd.org/mailman/listinfo/amber
> >>>> >
> >>>>
> >>>> --
> >>>> Dr. Vlad Cojocaru
> >>>> Max Planck Institute for Molecular Biomedicine
> >>>> Department of Cell and Developmental Biology
> >>>> Röntgenstrasse 20, 48149 Münster, Germany
> >>>> Tel: +49-251-70365-324; Fax: +49-251-70365-399
> >>>> Email: vlad.cojocaru[at]mpi-muenster.mpg.de
> >>>>
> >>>>
> >>>> _______________________________________________
> >>>> 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
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jul 16 2014 - 09:00:03 PDT
Custom Search