[gmx-users] Few queries about microsecond long simulations

[Felipe Merino]

Hi Anu,

Again that is up to you. I do not save a full precision file for my 
analysis. In the infinite disk space (and time!) limit of course you 
would like to have veolcities, forces, and positions of all atoms very 
frequently. However, if you do not foresee any analysis requiring 
velocities or high precision positions then you should be fine with the 
xtc file.

The xtc file is lighter and therefore you can save more info into it. 
Besides, you can write specific groups into it.

Felipe

On 19/08/15 11:27, anu chandra wrote:
> Hi Felipe and Ganesh,
>
> Do we really need to save coordinates in both of these two formats - trr
> and .xtc?. If one save the .trr file, what is the need of .xtc file since
> .trr can provide all the informations required with higher precision. All I
> am looking for is to save some disk space.
>
> Thanks
> Anu
>
> On Tue, Aug 18, 2015 at 3:19 PM, anu chandra <anu80125 at gmail.com> wrote:
>
>> Many thanks for your suggestions.
>>
>> On Tue, Aug 18, 2015 at 12:52 PM, Ganesh Shahane <ganesh7shahane at gmail.com
>>> wrote:
>>> Hi Anu,
>>>
>>> 1) It really depends on what kind of analysis you are doing. If you are,
>>> for example, studying bio-molecular interactions that take place over
>>> microsecond long timescales then saving trajectories every 200-500 ps is
>>> advisable. I myself do receptor-membrane atomistic simulations with a
>>> saving frequency of 500 ps. You don't really need saving which is as
>>> frequent as 10 ps.
>>>
>>> 2), 3) Whereas .xtc trajectories contain only the coordinates, .trr
>>> trajectories contain both coordinates as well as velocities. So unless you
>>> are going to do any analysis which involves the velocities of
>>> atoms/particles, its safe to not write-out the .trr file. Still you are
>>> never really sure if you are going to require velocities or not and so
>>> it's
>>> best to write out the .trr file as well.
>>>
>>> On Tue, Aug 18, 2015 at 4:30 PM, Felipe Merino <
>>> felipe.merino at mpi-dortmund.mpg.de> wrote:
>>>
>>>> Hi,
>>>>
>>>> This are all questions that mostly only you can answer.
>>>> On 18/08/15 12:42, anu chandra wrote:
>>>>
>>>>> Dear Gromcas users,
>>>>>
>>>>> I am planning to do microsecond long simulation of membrane-protein
>>> system
>>>>> and I have ~100,000 atoms in the systems. Since I am trying for
>>>>> microsecond
>>>>> long simulations and I have a quiet large system, it can generate large
>>>>> amount of data and in this regard I would like to clarify few queries
>>>>> before continue the simulations. My queries are,
>>>>>
>>>>> 1. How often I should save the trajectories? In general, I have seen
>>> that
>>>>> mostly use 2ps interval for saving trajectories, as shown below. Will
>>> it
>>>>> be
>>>>> safe if I use 10ps ( 5000 steps ) interval for saving trajectories?
>>>>>
>>>>> nstxout         = 1000          ; save coordinates every 2 ps
>>>>> nstvout         = 1000          ; save velocities every 2 ps
>>>>> nstenergy       = 1000          ; save energies every 2 ps
>>>>> nstlog          = 1000          ; update log file every 2 ps
>>>>>
>>>> The files you create and the frequency with which you write them depends
>>>> only on what you want to measure in the end. In principle, you want to
>>> save
>>>> every couple of decorrelation times, for whatever property interests
>>> you. I
>>>> assume that since you want a microsecond-long simulation then you are
>>> going
>>>> for a slow process and therefore you don't need so frequent saving.
>>> Just as
>>>> a comparison, for one of my simulations (~550k atoms, 1 µs) I save the
>>>> coordinates every 40 ps which gives me a final xtc of ~50Gb of space.
>>>>
>>>>> 2. How safe if I just write only the compressed .xtc file of the
>>>>> trajectories and not the .trr file during the production run? If its
>>> safe,
>>>>> how can I tell mdrun not to write the trr file?
>>>>> 3. Do I really need the trr file for any MD analysis?
>>>>>
>>>>> Again, you should know if you need the velocities later, or if need a
>>>> high precision for the coordinates. If you don't, then leave the mdp as,
>>>> nstxout         = 0
>>>> nstvout         = 0
>>>> nstenergy       = 1000
>>>> nstlog          = 1000
>>>> nstxtcout   = 1000
>>>>
>>>>> Many thanks in advance
>>>>>
>>>>> Anu
>>>>>
>>>> --
>>>> Felipe Merino
>>>> Max Planck Institute for Molecular Physiology
>>>> Department of Structural Biochemistry
>>>> Otto-Hahn-Str. 11
>>>> 44227 Dortmund
>>>> Phone: +49 231 133 2306
>>>>
>>>>
>>>> --
>>>> Gromacs Users mailing list
>>>>
>>>> * Please search the archive at
>>>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>>>> posting!
>>>>
>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>
>>>> * For (un)subscribe requests visit
>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>>> send a mail to gmx-users-request at gromacs.org.
>>>>
>>>
>>>
>>> --
>>> Best Regards,
>>> Ganesh Shahane
>>> --
>>> Gromacs Users mailing list
>>>
>>> * Please search the archive at
>>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>>> posting!
>>>
>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>
>>> * For (un)subscribe requests visit
>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>> send a mail to gmx-users-request at gromacs.org.
>>>
>>

-- 
Felipe Merino
Max Planck Institute for Molecular Physiology
Department of Structural Biochemistry
Otto-Hahn-Str. 11
44227 Dortmund
Phone: +49 231 133 2306