Could dt=0.004 be a factor in a blowup, and would that slow down the
calcs, I wonder. If I were to just try something, it would be dt=0.002.
Beyond that. I suggest saving coords for every step for a short run and
visualize, maybe compare the speed of the two molecules on CPU.
Bill
On 11/25/20 9:21 AM, Carlos Simmerling wrote:
> sorry but I don't really follow what you're seeing. When you mention the
> volume, do you mean the volume reported in the mdout? Or is it from
> visualizing? Did you image the system?
> something sounds wrong with the wallclock time requirement, are the
> energies ok? if it seems ok I'm not sure why it would be so slow, could it
> be a GPU problem (temperature, multiple jobs, etc)?
>
> On Wed, Nov 25, 2020 at 12:16 PM Amit Sharma (Asstt. Prof., MCARS) <
> asharma4.jmi.ac.in> wrote:
>
>> Dear Carlos,
>>
>> Thanks for suggesting Dan Roe's protocol. I will try that. But, for now I
>> equilibrated the dark state system (2v1a) exactly as per the method
>> described in your paper. What I find is, water and ions still move a quite
>> away from the protein during equilibration and the overall system has
>> expanded a lot in volume. Although the production MD seems to be going
>> fine, the issue now is that a 4ns simulation ( nstlim = 1000000, dt =
>> 0.004) is running for about 20 mins on the GPU. Whereas, the light state
>> system with the same size (2v1b, in which the system didn't expand much
>> during equilibration) took only 2 mins for a 4ns production run.
>>
>> Kindly let me know if this expansion (and the corresponding slow down) is a
>> usual thing to happen with the dark system. If not, can you please suggest
>> what measures I might have missed which make my run slower.
>>
>> Thanks,
>>
>> Amit
>>
>>
>> On Wed, Nov 25, 2020 at 8:50 PM Carlos Simmerling <
>> carlos.simmerling.gmail.com> wrote:
>>
>>> agreed - this the one I meant earlier in the thread when I said " Dan Roe
>>> also has a nice paper on equilibration"
>>>
>>> On Wed, Nov 25, 2020 at 10:12 AM Daniel Roe <daniel.r.roe.gmail.com>
>>> wrote:
>>>
>>>> Hi,
>>>>
>>>> You may also want to try the system preparation protocol outlined
>>>> here: https://aip.scitation.org/doi/abs/10.1063/5.0013849
>>>>
>>>> -Dan
>>>>
>>>> On Tue, Nov 24, 2020 at 7:41 AM Carlos Simmerling
>>>> <carlos.simmerling.gmail.com> wrote:
>>>>> it's quite different equilibration from what we used. I don't know if
>>> the
>>>>> problems you are seeing are due to that, but since the 2v1b works and
>>>> 2v1a
>>>>> does not, it suggests that it might be due to equilibration. When
>> using
>>>>> 2v1a as the initial structure, you are far from equilibrium at the
>>> start
>>>>> and need to slowly adjust to introducing the new bond and adjusting
>> to
>>>> new
>>>>> geometries in the flavin, etc. In my experience the GPU code can run
>>> into
>>>>> trouble if the system is at all unstable. You might try tracking your
>>>>> energies and other properties from the mdout files and see if they
>> have
>>>>> reached a plateau before going to the next step, your MD stages are
>>>> fairly
>>>>> short.
>>>>> Here is what we used that worked well for us:
>>>>>
>>>>> Minimization and equilibration were carried out using Amber version
>>>>> 16.(39,40) Initial minimization was performed for 100 000 steps with
>>>>> restraints on all atoms except hydrogens, water, C450, and the
>> flavin.
>>>> The
>>>>> restraint force constant was 100 kcal·mol–1·Å–2. Next, the system was
>>>>> heated from 100 to 298 K over 1 ns in the NVT ensemble, with a time
>>> step
>>>> of
>>>>> 1 fs and SHAKE on all bonds including hydrogen. A nonbonded cutoff of
>>> 8 Å
>>>>> was used, with long-range electrostatics included by the particle
>> mesh
>>>>> Ewald method.(41) The same restraints were maintained. Temperature
>> was
>>>>> maintained using a Langevin thermostat with γ set to 1.0. Next,
>>> pressure
>>>>> and density were relaxed using 1 ns NPT simulation at 298 K with a
>>> strong
>>>>> pressure coupling constant of 0.1 and all other parameters maintained
>>>> from
>>>>> the prior step. Next, 1 ns MD was performed using the same protocol
>> but
>>>>> with restraint force constant reduced to 10.0 and pressure coupling
>>>>> constant increased to 0.5. Next, minimization was performed for 10000
>>>> steps
>>>>> after removing all restraints except those on protein backbone CA, N,
>>>> and C
>>>>> atoms, and no restraints on C450. Next, 1 ns MD in the NPT ensemble
>> at
>>>> 298
>>>>> K was performed using the same protocol as above, with restraints
>> only
>>> on
>>>>> backbone CA, N, and C atoms excepting C450. The restraint force
>>> constant
>>>>> was then reduced from 10.0 to 1.0 and an additional 1 ns MD was
>> carried
>>>>> out, followed by 1 ns MD with restraint force constant reduced to
>> 0.1.
>>>>> Finally, 1 ns of fully unrestrained MD was performed with NPT at 298
>> K.
>>>>> On Tue, Nov 24, 2020 at 6:44 AM Amit Sharma (Asstt. Prof., MCARS) <
>>>>> asharma4.jmi.ac.in> wrote:
>>>>>
>>>>>> I tried simulation using 2v1b (crystal structure in photoexcited
>>>> state). It
>>>>>> works fine in that case and is currently running at 2 micro second.
>>>>>>
>>>>>> My equilibration protocol is as follows
>>>>>>
>>>>>> 1. 3000 cycle minimization under 100 kcal mol-1 A-2 restraint on
>> the
>>>>>> protein
>>>>>> 2. 5000 cycle minimization under 25 kcal mol-1 A-2 restraint on the
>>>> protein
>>>>>> 3. 100000 cycle minimization without any restraints
>>>>>> 4. 20ps md at nvt where system was heated from 0 to 300K, with 10
>>> kcal
>>>>>> mol-1 A-2 restraint on the protein, 2fs time step by keeping SHAKE
>>> on.
>>>> cut
>>>>>> of 10 A
>>>>>> 5. 80ps md at nvt at 300 K, with 0.5 kcal mol-1 A-2 restraint on
>> the
>>>>>> protein, 2fs time step by keeping SHAKE on. cut of 10 A
>>>>>>
>>>>>> System remains fine until step 5, but blows when I start (using
>> gpu)
>>>> ntp
>>>>>> (ntb = 2, pressure 1atm, ntp = 1) at 300K.
>>>>>>
>>>>>>
>>>>>> When doing the same using cpu, it seems fine at least for initial
>>>> 100ps.
>>>>>> But, issue is that I cannot go on cpu for the simulation where a
>> few
>>>> micro
>>>>>> second run is needed before we could actually see J alpha
>> unfolding.
>>>>>> In your paper you retain the crystallographic water and use four
>>> point
>>>> opc
>>>>>> water. But, what I am using is tip3p. Can that make a difference?
>>>>>>
>>>>>> Amit
>>>>>>
>>>>>>
>>>>>> On Tue, Nov 24, 2020 at 4:40 PM Carlos Simmerling <
>>>>>> carlos.simmerling.gmail.com> wrote:
>>>>>>
>>>>>>> Have you tried the simulation without adding the adduct? Also you
>>>> didn't
>>>>>>> describe your equilibration protocol. I believe the one we used
>>>> should be
>>>>>>> described in our paper on that system. Dan Roe also has a nice
>>> paper
>>>> on
>>>>>>> equilibration.
>>>>>>> Carlos
>>>>>>>
>>>>>>> On Tue, Nov 24, 2020, 6:05 AM Amit Sharma (Asstt. Prof., MCARS) <
>>>>>>> asharma4.jmi.ac.in> wrote:
>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I trying to simulate protein having of FMN Cys adduct.
>>>>>>>>
>>>>>>>> When running md using nvt and positional restraints of 0.5 kcal
>>>> mol-1
>>>>>> A-2
>>>>>>>> on the protein, the system remains fine.
>>>>>>>>
>>>>>>>> But as I switch to npt and remove all the restraints system
>> blows
>>>> up
>>>>>> and
>>>>>>>> water molecules start flying away.
>>>>>>>>
>>>>>>>> Error message is as follows:
>>>>>>>>
>>>>>>>> "ERROR: Calculation halted. Periodic box dimensions have
>> changed
>>>> too
>>>>>>> much
>>>>>>>> from their initial values.
>>>>>>>> Your system density has likely changed by a large amount,
>>>> probably
>>>>>> from
>>>>>>>> starting the simulation from a structure a long way from
>>>> equilibrium.
>>>>>>>> [Although this error can also occur if the simulation has
>> blown
>>>> up
>>>>>> for
>>>>>>>> some reason]
>>>>>>>>
>>>>>>>> The GPU code does not automatically reorganize grid cells and
>>>> thus
>>>>>> you
>>>>>>>> will need to restart the calculation from the previous
>> restart
>>>> file.
>>>>>>>> This will generate new grid cells and allow the calculation
>> to
>>>>>>> continue.
>>>>>>>> It may be necessary to repeat this restarting multiple times
>> if
>>>> your
>>>>>>>> system
>>>>>>>> is a long way from an equilibrated density.
>>>>>>>>
>>>>>>>> Alternatively you can run with the CPU code until the density
>>> has
>>>>>>>> converged
>>>>>>>> and then switch back to the GPU code."
>>>>>>>>
>>>>>>>> Can anyone suggest how to deal with this?
>>>>>>>>
>>>>>>>> Thank You,
>>>>>>>>
>>>>>>>> Amit
>>>>>>>>
>>>>>>>> --
>>>>>>>> _
>>>>>>>> _
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> *Please consider the environment before printing this email.*
>>>>>>>> The
>>>>>>>> information contained in this electronic message and any
>>>> *attachments
>>>>>> to
>>>>>>>> this message are intended for the exclusive use of addressee(s)
>>>> and may
>>>>>>>> contain proprietary, confidential or privileged information. If
>>>> you are
>>>>>>>> not
>>>>>>>> the intended recipient, you should not disseminate, distribute
>>> **or
>>>>>> copy
>>>>>>>> this e-mail. Please notify the
>>>>>>>> sender immediately and destroy all copies of
>>>>>>>> this message and any attachments.***
>>>>>>>> **
>>>>>>>> **WARNING***: The recipient should
>>>>>>>> check this email and any attachments for
>>>>>>>> the presence of viruses. JMI
>>>>>>>> accepts no liability for any damage caused by any
>>>>>>>> virus transmitted by this
>>>>>>>> email.*
>>>>>>>> _______________________________________________
>>>>>>>> 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
>>>>>>>
>>>>>> --
>>>>>> _
>>>>>> _
>>>>>>
>>>>>>
>>>>>>
>>>>>> *Please consider the environment before printing this email.*
>>>>>> The
>>>>>> information contained in this electronic message and any
>> *attachments
>>>> to
>>>>>> this message are intended for the exclusive use of addressee(s) and
>>> may
>>>>>> contain proprietary, confidential or privileged information. If you
>>> are
>>>>>> not
>>>>>> the intended recipient, you should not disseminate, distribute **or
>>>> copy
>>>>>> this e-mail. Please notify the
>>>>>> sender immediately and destroy all copies of
>>>>>> this message and any attachments.***
>>>>>> **
>>>>>> **WARNING***: The recipient should
>>>>>> check this email and any attachments for
>>>>>> the presence of viruses. JMI
>>>>>> accepts no liability for any damage caused by any
>>>>>> virus transmitted by this
>>>>>> email.*
>>>>>> _______________________________________________
>>>>>> 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
>>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>> --
>> _
>> _
>>
>>
>>
>> *Please consider the environment before printing this email.*
>> The
>> information contained in this electronic message and any *attachments to
>> this message are intended for the exclusive use of addressee(s) and may
>> contain proprietary, confidential or privileged information. If you are
>> not
>> the intended recipient, you should not disseminate, distribute **or copy
>> this e-mail. Please notify the
>> sender immediately and destroy all copies of
>> this message and any attachments.***
>> **
>> **WARNING***: The recipient should
>> check this email and any attachments for
>> the presence of viruses. JMI
>> accepts no liability for any damage caused by any
>> virus transmitted by this
>> email.*
>> _______________________________________________
>> 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
>
--
Phobrain.com
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 25 2020 - 18:30:02 PST
