Chris,
adding to Hai's post, I'd also suggest to use a chain-job script that
checks for your criteria and does a resubmit only if all conditions are
met.
With this, you'd maybe create more single trajectory files, but you
could set in advance, how much computational time you would waste in the
worst case; and you'd have all files in defined states, since all
calculations are ended via Amber, not via kill, qdel, or scancel.
IMHO, a direct mechanism, which allowed to flag a running Amber job a
premature stop condition and would act similar to nstlim had been
reached, does not exist.
Best,
Anselm
Bioinformatik | NHR.FAU
Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)
Germany
On 03/17/2022 05:10 PM, Hai Nguyen wrote:
> Hi
>
> You can write your own cron script to calculate rmsd and very 15 minute. If
> the rmsd satisfies your criteria, the script can kill the job (depends on
> how you submit the job, “kill”, “qdel” and so on.
>
> Hai
>
> On Wed, Mar 16, 2022 at 3:32 PM Chris Lee <chrishankyu.lee.mail.utoronto.ca>
> wrote:
>
>> Hi all,
>>
>> I'm wondering if there is a way to 'monitor' and pre-emptively stop or
>> cancel a MD run if certain criteria are met. Right now I want to look at if
>> the RMSD value goes above a certain threshold, but I would be interested to
>> know if there is a general method to do this with other metrics of interest.
>>
>>
>> I need to run a huge number of MDs (on the order of 10,000s) and ideally I
>> would like to run all of them on the order of 10s of nanoseconds; but
>> obviously this will take a huge amount of computational resources and take
>> forever.
>> I have good reason to believe most of these starting structures will have
>> their RMSD values shoot up high very early on in the simulation, so I want
>> to filter these out and not waste computational time.
>>
>> Thanks,
>> Chris
>> _______________________________________________
>> 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 Thu Mar 17 2022 - 10:30:02 PDT
