Jason,
As I’ve come across similar situations, I’d like to chip in.
In situations like this, would it help to do (i) clustering of the trajectory, and (ii) running MMPBSY.py on each cluster?
Regards
George
On 20 Apr 2015, at 15:09, Jason Swails <jason.swails.gmail.com> wrote:
>
>> On Apr 20, 2015, at 4:41 AM, windy <chxp_moon.163.com> wrote:
>>
>> Hi, everyone,
>>
>> My substrate has a conformation change during MD, but the time span of each conformation is so different.
>>
>> For example, conformation I spans 30ns , conformation II 10 ns , conformation III 2.5 ns, conformation IV 20 ns.
>>
>> SO How to set interval or the start/end frames when enery calculation in MMPBSY.py? What should I consider first ? The frames each conformation? The time span (may be 2.5 ns) must same ? Or maybe just free to choose calculation frames or time spans (start/end frame), the only thing is that you should keep the conformations in each calculation are stable?
>>
>> Anyone have good ideas? Thanks for you reply .
>
> What you are effectively asking is how do you properly subsample from a trajectory. The naive approach provided by MM/PBSA scripts is to sample every N’th frame of a trajectory, where the interval N is chosen such that you calculate energies for the desired number of snapshots. If you have 20,000 total frames and you want to analyze only 200 of them, then you would set your interval to 100.
>
> Another way of subsampling a larger sample is to take 200 *random* snapshots from your entire trajectory, although I’ve never really seen a problem with the approach above. What you typically *shouldn’t* do is to use prior knowledge about the trajectory in order to sample non-uniformly from the trajectory. For example, taking an equal number of snapshots from each of the 4 conformations will make conformation III contribute as much to the energies as conformation IV and conformation I; despite the fact that it contributes only ~4% of the total ensemble compared to 50% for conformation I (or 33% for conformation IV).
>
> I emphasize the typically in my above statement, because obviously what you should do depends on what it is you want to learn.
>
> HTH,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
>
> _______________________________________________
> 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 Mon Apr 20 2015 - 11:30:02 PDT
