[gmx-users] how can we obtain replicated trajectory?

[Mark Abraham]

On Fri, Mar 22, 2013 at 3:54 PM, Albert <mailmd2011 at gmail.com> wrote:

> On 03/22/2013 03:45 PM, Mark Abraham wrote:
>
>> As you can see in your plot, no replicas from 400K got down to 300K in the
>> 1.5ns you simulated. So you've so far derived no benefit from REMD (unless
>> you want sampling at lots of temperatures). Perhaps a simulation a hundred
>> times longer might get a few such events? You might as well have run a
>> simulation at 400K to see what it did... What do you hope REMD will help
>> you observe?
>>
>> Mark
>>
>
>
> HI Mark
>
>  thanks a lot for such kind advices and comments.
>
>  There is a ligand in my protein binding site and I hope REMD can enable
> me how the ligand escape binding pocket coupled with some big loop movement.
>

"How" implies you need to see the process by which such an event occurs,
which you can't do in REMD because if the trajectory is continuous then its
ensemble was not, and vice versa. If your REMD sampling is exhaustive, then
you may "catch" some frames around 300K in the process of escaping, but
probably you can't afford that much computer time. Using a pulling
simulation to force the reaction coordinate to sample the space that
interests you is likely a much more productive way to enhance your
sampling. Check the literature!

Mark