A lot of this question is tied up in what you consider to be an independent
run. For me, an independent run must start from an independent
conformational sample, which is unlikely within 20 ns for much beyond a box
of water. When I do independent runs, I like to randomize everything that
is not completely known going in. Therefore, for a simulation of a protein
in a box of water, given that all simulations will start from the same
crystal structure, I like to repeat all initial structure generation
procedures using a different random seed each time. So I would in this way
randomize things like side chain building for chains that were incomplete
in the crystal structure, initial placement of waters during hydration,
initial placement of ions, etc. You might think that initial hydration
relaxes quickly, and in many cases it will, but interfacial or internal
waters may have long relaxation times and if you happened to put one water
in one place that relaxes slowly (could easily be >20 ns) then your repeats
will not be as independent as you think they are. For membrane proteins, I
build a neat lipid bilayer and simulate it for a while and then use
different configurations of the bilayer to separately insert the protein in
each repeat. If in doubt, randomize!
On Wed, Oct 4, 2017 at 10:58 PM, Elvis Martis <elvis.martis.bcp.edu.in>
wrote:
> Hi,
> I would like to know if I run a 20ns MD simulation and save the trajectory
> every 1ns, and so I have 20 separate trajectories (each is 1ns). Are these
> 20 trajectories consider as independent MD run?
> >>>> These cannot be considered independent trajectories because you are
> merely saving them and not restarting your simulations (using ig=-1).
> Instead, after equilibrations, you can run, as many as you wish, parallel
> simulations using ig=-1 and these will be considered independent
> trajectories. Even better if you start from slightly different starting
> points(conformations).
>
> Second scenario:
>
> If I extracted several structures from a single 20ns trajectory (say
> structure at 5ns and 10ns) and use these structures to run another two 20ns
> MD simulations (same parameter and setting as the previous 20ns
> simulations), are these two 20ns simulations using structure at 5ns and
> 10ns as starting structure consider as independent MD run?
> >> Yes here you can that.
>
> Best Regards
> Elvis Martis
> Mumbai, INDIA.
>
> ________________________________________
> From: colvin <colvin4367.gmail.com>
> Sent: 05 October 2017 08:41
> To: AMBER Mailing List
> Subject: [AMBER] Independent MD runs
>
> Dear all.
>
> I would like to know if I run a 20ns MD simulation and save the trajectory
> every 1ns, and so I have 20 separate trajectories (each is 1ns). Are these
> 20 trajectories consider as independent MD run?
>
> Second scenario:
>
> If I extracted several structures from a single 20ns trajectory (say
> structure at 5ns and 10ns) and use these structures to run another two 20ns
> MD simulations (same parameter and setting as the previous 20ns
> simulations), are these two 20ns simulations using structure at 5ns and
> 10ns as starting structure consider as independent MD run?
>
> If not, what should I do to run an independent MD run?
>
> Thank you!
> _______________________________________________
> 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 Oct 04 2017 - 23:00:03 PDT
