Re: Cavernous Depressions in Water Box

From: Oscar Bastidas (
Date: Thu Dec 31 2020 - 15:26:31 CST

Hi Peter,

Thank you so much for the diagnosis. I'll set up my subsequent runs
accordingly, as well as keep the mailing list on this and future threads.
Happy New Year.


Oscar Bastidas, Ph.D.
Postdoctoral Research Associate
University of Minnesota

On Thu, Dec 31, 2020, 3:05 PM Peter Freddolino <> wrote:

> Dear Oscar,
> Your answer here reveals the problem. You said awhile ago that you were
> running an NPT simulation, but if you don't have pressure controls defined
> in your config file, that is not correct. It sounds like you are actually
> running an NVT simulation, which explains all of the symptoms that you
> describe. Your box is under-solvated, and in the absence of the ability for
> the volume to shrink, you instead get cavity formation. This should be
> fixed by doing a proper NPT simulation.
> Also, please keep the mailing list copied on any reply threads.
> Best,
> Peter
> On Thu, Dec 31, 2020 at 3:37 AM Oscar Bastidas <> wrote:
>> Hello Peter,
>> My apologies for a delayed response to your previous e-mail to me in
>> regards to my water box indentation issues I had corresponded with you on
>> earlier. I would like to get your diagnoses on what might be causing the
>> indentations/depressions I am observing.
>> To address your most previous response regarding what kind of pressure
>> controls I was using in my configuration file, my response is that I am not
>> using any of the namd-defined pressure controls (useGroupPressure,
>> useFlexibleCell, useConstantRatio, etc...) in my configuration file. I am,
>> however, using the Langevin dynamics for temperature control.
>> Below is the snapshot of that temperature control section:
>> # Constant Temperature Control
>> langevin on
>> langevinDamping 1.0
>> langevinTemp $temp
>> langevinHydrogen off
>> Thanks.
>> Oscar
>> Oscar Bastidas, Ph.D.
>> Postdoctoral Research Associate
>> University of Minnesota
>> On Nov 28, 2020 6:07 PM, "Peter Freddolino" <> wrote:
>> Hi Oscar,
>> This still seems a bit odd. Could you please send me the parts of your
>> config file related to pressure control?
>> Thanks,
>> Peter
>> On Fri, Nov 27, 2020 at 12:49 AM Oscar Bastidas <> wrote:
>>> Hi Peter,
>>> Upon looking at my unit cell trajectory, my unit cell parameters, over
>>> the course of the simulation experience the following trends: x and y-basis
>>> vectors of the unit cell decrease over time, z-basis vectors increase and
>>> overall unit cell volume decreases. I observed these trends by just
>>> plotting all my basis vectors and volumes in chronological order and doing
>>> a simple linear curve fit on the data. I should note that the slopes are
>>> very small numbers though, no dramatic changes, with the orders of
>>> magnitude being either -0.000X for negative slopes or +0.000X for the
>>> single positive slope. Lastly, R squared values for all basis vectors are
>>> also small.
>>> Do the above observartions correspond to a correctly working barostat?
>>> Hopefully this sheds some insight into how I can re-run the simulations to
>>> avoid the depression/cavern. Thanks.
>>> Oscar
>>> Oscar Bastidas, Ph.D.
>>> Postdoctoral Research Associate
>>> University of Minnesota
>>> On Sun, Nov 22, 2020, 11:56 PM Peter Freddolino <>
>>> wrote:
>>>> I think I was a bit unclear in what I'm asking for - actually I am
>>>> curious about your unit cell parameters themselves, and whether they
>>>> stabilize over time or are still changing (presumably shrinking) as the
>>>> simulation goes on. If there is a cavity forming they really should shrink
>>>> to compensate, if your barostat is working.
>>>> Best,
>>>> Peter
>>>> On Mon, Nov 23, 2020 at 12:52 AM Oscar Bastidas <>
>>>> wrote:
>>>>> Hi Peter,
>>>>> Yes, I will definitely send you my pressure conf file parameters.
>>>>> I will also take a look at my unit cell trajectories again to see if
>>>>> the depression evolved gradually or if it was there from the get go and let
>>>>> you know what I find. Thanks.
>>>>> Oscar
>>>>> Oscar Bastidas, Ph.D.
>>>>> Postdoctoral Research Associate
>>>>> University of Minnesota
>>>>> On Sun, Nov 22, 2020, 11:41 PM Peter Freddolino <>
>>>>> wrote:
>>>>>> Hi Oscar,
>>>>>> Could you please send me the parameters in your config file
>>>>>> pertaining to pressure control? That is, the barostat itself and extended
>>>>>> system definition. Also, if you look at a trace of your unit cell
>>>>>> dimensions vs time, do they look like they are stabilizing or are they
>>>>>> still changing at the end of your simulation?
>>>>>> Thanks,
>>>>>> Peter
>>>>>> On Sun, Nov 22, 2020 at 1:29 PM Oscar Bastidas <>
>>>>>> wrote:
>>>>>>> Dear Peter,
>>>>>>> Here are the answers to the questions:
>>>>>>> 1) How did I solvate? I used CHARMM-GUI for TIP3P water molecule
>>>>>>> explict hydration.
>>>>>>> 2) What ensemble am I running in? Everything I've been doing is NPT
>>>>>>> and I do not have any constraints on my system.
>>>>>>> 3) Any other simulation features being used besides standard MD?
>>>>>>> No, I am doing standard MD.
>>>>>>> Hopefully this can narrow down the diagnosis. Please let me know of
>>>>>>> any thoughts. Thanks.
>>>>>>> Oscar
>>>>>>> Oscar Bastidas, Ph.D.
>>>>>>> Postdoctoral Research Associate
>>>>>>> University of Minnesota
>>>>>>> On Sat, Nov 21, 2020, 9:14 PM Peter Freddolino <>
>>>>>>> wrote:
>>>>>>>> Dear Oscar,
>>>>>>>> an important general question to ask yourself is, does this
>>>>>>>> simulation behavior match what you'd expect in the real system? Almost
>>>>>>>> assuredly not, which means it is a problem that needs to be addressed. As
>>>>>>>> you've inferred, in general it seems that your simulation doesn't have
>>>>>>>> enough solvent given all other characteristics of the system. We would need
>>>>>>>> a bit more information to give good advice though, notably including:
>>>>>>>> -how did you solvate the system?
>>>>>>>> -what ensemble are you running in? (this is crucial. A constant
>>>>>>>> pressure simulation should not allow the formation of the type of cavity
>>>>>>>> that you describe unless there are some constraints on the system or there
>>>>>>>> is a region of the system that is topologically isolated from the rest --
>>>>>>>> you can see behavior like this, for example if you don't sufficiently
>>>>>>>> solvate the interior of a simulation containing a viral capsid)
>>>>>>>> -are there any other simulation features being used besides a
>>>>>>>> standard md simulation?
>>>>>>>> Best,
>>>>>>>> Peter
>>>>>>>> On Sat, Nov 21, 2020 at 9:08 PM Oscar Bastidas <>
>>>>>>>> wrote:
>>>>>>>>> Hello,
>>>>>>>>> I've recently run a high salt (3X physiological conditions) NAMD
>>>>>>>>> simulation on a large protein and the resultant production run trajectories
>>>>>>>>> show these cavernous depressions on the water box - two bowl-like cavities,
>>>>>>>>> one cavity on one face of the cube and the other on the opposite face.
>>>>>>>>> I've looked up the phenomenon in the archived mailing list and no
>>>>>>>>> one there has mentioned a high salt scenario like with what I'm working
>>>>>>>>> with here so I'm not sure if my cause for the water box cavities is the
>>>>>>>>> same cause for the archived event.
>>>>>>>>> My collaborators and I are trying to decide how to remedy this
>>>>>>>>> whether we should:
>>>>>>>>> 1) Redo the simulation while fixing the protein's atoms so they
>>>>>>>>> don't move and letting the waters fill in the cavities under a high
>>>>>>>>> temperature series of NVT equilibration runs or if
>>>>>>>>> 2) We should use the grand canonical ensemble and add waters
>>>>>>>>> according to the chemical potential if NAMD has an option for doing this.
>>>>>>>>> Does anyone have any thoughts as to what happened to my high salt
>>>>>>>>> system here and how we can address the cavities situation? Do the
>>>>>>>>> simulation results for such cavities typically affect the simulation in
>>>>>>>>> aberrant ways that wouldn't happen if the cavities weren't there?
>>>>>>>>> Thanks.
>>>>>>>>> Oscar B.
>>>>>>>>> Oscar Bastidas, Ph.D.
>>>>>>>>> Postdoctoral Research Associate
>>>>>>>>> University of Minnesota

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2020 - 23:17:15 CST