Hi,
On Thu, Nov 3, 2011 at 3:11 PM, Sarah Edwards <sarah.sarahjedwards.com> wrote:
> Hello all!
> 1. Do LMOD simulations need to be as long (i.e. as many steps) as standard
> MD?
This is really an apples to oranges kind of thing. MD and LMOD are
doing very different things, and you really can't compare an MD step
to an LMOD step. That said, you will generally run less LMOD steps
than you would for MD, but that's just because an LMOD step involves a
lot more (follow mode uphill, cross barrier, find new minimum) than an
MD step. As far as I know there is no hard and fast rule for
determining the number of LMOD steps you have to run. It's the same as
MD in that respect - you have to run as many iterations as necessary
to observe whatever behavior you're interested in. For very small
local optimizations I've gotten decent results from running < 10
iterations of LMOD, but for larger systems (hundreds or thousands of
atoms) you will likely need to run far more iterations. Note that your
results are also going to be sensitive to the number of modes you're
calculating and following each iteration.
> 2. Is it physically reasonable to use NMR-type restraints during an LMOD
> simulation, or can that lead to non-physical results?
This I don't really have experience with but I can try to offer an
educated guess. It probably depends on what exactly you're restraining
and how you're restraining it. In Sander I'm pretty sure the NMR
restraints aren't applied during the LMOD procedure itself, but in
between calls to the LMOD routine (others please correct me if I'm
wrong). So for example if you're attempting to restrain a distance
between two non-bonded atoms and after an LMOD iteration your
structure has those atoms far apart that could be bad. If the intent
of restraints is apply LMOD to only a subset of your system you could
also use LMOD as it is implemented in NAB, which allows you to easily
freeze (i.e. no forces are applied to) regions of your system.
-Dan
>
> Thanks!
> ~Sarah
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
-------------------------
Daniel R. Roe, PhD
Postdoctoral Associate
BioMaPS Institute, Rutgers University
610 Taylor Road
Piscataway, NJ 08854
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Nov 03 2011 - 15:00:04 PDT
