[gmx-users] RMSF started increasing after 60 ns simulation

Wed Apr 11 16:47:04 CEST 2018

On 4/11/18 10:44 AM, MD wrote:
> Thank you Justin :)
> I wonder what is your strategy when considering what criteria could be used
> for determining convergence of the simulation?

If all the relevant quantities that I care about are invariant over time 
based on error estimates, it's converged.

-Justin

> Ming
>
> On Wed, Apr 11, 2018 at 10:30 AM, Justin Lemkul <jalemkul at vt.edu> wrote:
>
>>
>> On 4/11/18 10:26 AM, MD wrote:
>>
>>> Thank you Justin. Two quick questions (learning questions :))
>>> 1. How do I apply several invocations of trjconv, (a quick guide would be
>>> appreciated)
>>>
>> http://www.gromacs.org/Documentation/Terminology/Periodic_
>> Boundary_Conditions#Suggested_trjconv_workflow
>>
>> If I had $0.10 for every time I posted that link, I could fund my summer
>> salary... :)
>>
>> 2. Several literature used only 90-100 ns time window for RMSF analysis in
>>> a 100ns run. Does that mean the later the time window is the more value of
>>> RMSF it has?
>>>
>> Whenever a given time frame has been chosen for a particular analysis, it
>> should be accompanied by an explanation of why that time frame was chosen.
>> Is it because the authors believed the simulation did not converge until 90
>> ns, and therefore they discarded 0-90 ns and only analyzed from 90-100 ns?
>> That may be the case, but for a slow-converging system, I would expect the
>> simulation to be run much longer to collect more data. 10 ns is not much
>> time at all for anything useful.
>>
>> RMSF is particularly sensitive to equilibration; the initial frames in
>> unrestrained MD are often not valid to include in this analysis, but one
>> should consider very carefully what criteria are being used for determining
>> convergence of the simulation. No one quantity will tell you this, even if
>> it is aesthetically pleasing.
>>
>> -Justin
>>
>> Thanks again,
>>> MD
>>>
>>> On Wed, Apr 11, 2018 at 10:17 AM, Justin Lemkul <jalemkul at vt.edu> wrote:
>>>
>>>
>>>> On 4/11/18 10:16 AM, MD wrote:
>>>>
>>>> Hi,
>>>>> I wouldn't do cherry-picking. I am just confused how I can trust the
>>>>> data
>>>>> if the RMSD goes crazy in that range? Also based on the command I used I
>>>>> think the PBC has been taken care of, please correct me if I am wrong :)
>>>>>
>>>>> And It would be very interesting, if the RMSF of wild type started
>>>>> fluctuating after 70 ns for real.
>>>>>
>>>>> Also, yes you are right, this is a single run and I will need to
>>>>> replicate
>>>>> the simulation for several times. At the mean time I  would really like
>>>>> to
>>>>> learn how I can make a judgement which ranges of data I could use for
>>>>> the
>>>>> analysis :)
>>>>>
>>>>> The best tool you have is your eyes. Watch the trajectory. Understand
>>>> what
>>>> is happening during the time points for which the data behave very
>>>> differently. Maybe something interesting is happening and you need to
>>>> consider it. Maybe your PBC treatment is inadequate (often you need
>>>> several
>>>> invocations of trjconv to get things right, very few cases allow you to
>>>> use
>>>> a single, magic command).
>>>>
>>>>
>>>> -Justin
>>>>
>>>> Thank you,
>>>>
>>>>> MD
>>>>>
>>>>> On Wed, Apr 11, 2018 at 10:08 AM, Justin Lemkul <jalemkul at vt.edu>
>>>>> wrote:
>>>>>
>>>>>
>>>>> On 4/11/18 10:04 AM, MD wrote:
>>>>>> Hi,
>>>>>>
>>>>>>> Yes I am comparing the trajectories side by side and I ran a RMSD of
>>>>>>> wild
>>>>>>> type too.
>>>>>>> I looked at the RMSD of wild type and it looks like it became unstable
>>>>>>> after 70ns. In this case the only approach is to compare the
>>>>>>> trajectories
>>>>>>> in time windows before 70ns. Would that be right?
>>>>>>>
>>>>>>> I'd argue that that's cherry-picking, taking preferable data and
>>>>>>>
>>>>>> ignoring
>>>>>> the real outcome. If the protein structure goes haywire (and it really,
>>>>>> truly isn't a PBC effect, which it is in about 9 out of 10 cases...)
>>>>>> then
>>>>>> you should investigate what's going on. Maybe the .mdp settings are
>>>>>> incorrect and causing the model physics to break down. Maybe the force
>>>>>> field you chose is not sufficiently accurate for simulating the system.
>>>>>> Is
>>>>>> the behavior reproducible across multiple simulations, or are you
>>>>>> looking
>>>>>> at an outlier?
>>>>>>
>>>>>>
>>>>>> -Justin
>>>>>>
>>>>>> Thank you,
>>>>>>
>>>>>> MD
>>>>>>>
>>>>>>> On Wed, Apr 11, 2018 at 9:42 AM, Justin Lemkul <jalemkul at vt.edu>
>>>>>>> wrote:
>>>>>>>
>>>>>>>
>>>>>>> On 4/11/18 9:41 AM, MD wrote:
>>>>>>>
>>>>>>>> Hi Justin,
>>>>>>>>
>>>>>>>> This was the command I used for both wild type and mutant. I wonder
>>>>>>>>> if
>>>>>>>>> you
>>>>>>>>> could think of any reason that made only the wild type have the PBC
>>>>>>>>> effects?
>>>>>>>>>
>>>>>>>>> gmx trjconv -s md.tpr -f md.xtc -o md_noPBC.xtc -pbc mol -ur compact
>>>>>>>>>
>>>>>>>>> Visualize each trajectory as Mark suggested.
>>>>>>>>>
>>>>>>>>> -Justin
>>>>>>>>
>>>>>>>> Thank you,
>>>>>>>>
>>>>>>>> MD
>>>>>>>>
>>>>>>>>> On Wed, Apr 11, 2018 at 9:23 AM, Justin Lemkul <jalemkul at vt.edu>
>>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On 4/11/18 9:22 AM, MD wrote:
>>>>>>>>>
>>>>>>>>> Hi Gromacs folks,
>>>>>>>>>> I was trying to use RMSF to compare c-alpha of a wild type and
>>>>>>>>>> mutant
>>>>>>>>>>
>>>>>>>>>>> in a
>>>>>>>>>>> 100 ns simulation. I discarded the 10-30 ns and compared them side
>>>>>>>>>>> by
>>>>>>>>>>> side
>>>>>>>>>>> in non-overlapping time windows of 30-40ns, 40-50ns, 50-60ns,
>>>>>>>>>>> 60-70
>>>>>>>>>>> ns,
>>>>>>>>>>> 70-80 ns, 80-90 ns and 90-100ns. The RMSF of wild type and mutant
>>>>>>>>>>> overlay
>>>>>>>>>>> pretty well with exceptions of some regions with minor flexibility
>>>>>>>>>>> differences, but the c-alpha RMSF of wild type went crazy starting
>>>>>>>>>>> from
>>>>>>>>>>> 70-80ns (7-8 folds of a more drastic change) and lasted until the
>>>>>>>>>>> end,
>>>>>>>>>>> while the mutant stayed in the same range of changing.
>>>>>>>>>>>
>>>>>>>>>>> Does this mean that I should trust the simulation after 70ns, or I
>>>>>>>>>>> should
>>>>>>>>>>> continue the simulation and see if the system will calm down
>>>>>>>>>>> eventually?
>>>>>>>>>>> Any help will be appreciated.
>>>>>>>>>>>
>>>>>>>>>>> It sounds to me like one simulation isn't properly corrected for
>>>>>>>>>>> PBC
>>>>>>>>>>>
>>>>>>>>>>> effects (protein jumping back and forth) while the other is. Make
>>>>>>>>>>>
>>>>>>>>>> sure
>>>>>>>>>> the
>>>>>>>>>> trajectory is sensible; gmx rmsf doesn't
>>>>>>>>>> <https://maps.google.com/?q=trajectory+is+sensible;+gmx+rmsf
>>>>>>>>>> +doesn't+&entry=gmail&source=g>intrinsically
>>>>>>>>>> account for PBC
>>>>>>>>>> so
>>>>>>>>>> you have to judiciously apply trjconv before running the analysis.
>>>>>>>>>>
>>>>>>>>>> -Justin
>>>>>>>>>>
>>>>>>>>>> --
>>>>>>>>>> ==================================================
>>>>>>>>>>
>>>>>>>>>> Justin A. Lemkul, Ph.D.
>>>>>>>>>> Assistant Professor
>>>>>>>>>> Virginia Tech Department of Biochemistry
>>>>>>>>>>
>>>>>>>>>> 303 Engel Hall
>>>>>>>>>> 340 West Campus Dr.
>>>>>>>>>> Blacksburg, VA 24061
>>>>>>>>>>
>>>>>>>>>> jalemkul at vt.edu | (540) 231-3129
>>>>>>>>>> http://www.thelemkullab.com
>>>>>>>>>>
>>>>>>>>>> ==================================================
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> --
>>>>>>>>>> Gromacs Users mailing list
>>>>>>>>>>
>>>>>>>>>> * Please search the archive at http://www.gromacs.org/Support
>>>>>>>>>> /Mailing_Lists/GMX-Users_List before pos
>>>>>>>>>> <https://maps.google.com/?q=/Mailing_Lists/GMX-Users_List+be
>>>>>>>>>> fore+pos&entry=gmail&source=g>
>>>>>>>>>>
>>>>>>>>>> ting!
>>>>>>>>>>
>>>>>>>>>> * 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.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> --
>>>>>>>>>>
>>>>>>>>>> ==================================================
>>>>>>>> Justin A. Lemkul, Ph.D.
>>>>>>>> Assistant Professor
>>>>>>>> Virginia Tech Department of Biochemistry
>>>>>>>>
>>>>>>>> 303 Engel Hall
>>>>>>>> 340 West Campus Dr.
>>>>>>>> Blacksburg, VA 24061
>>>>>>>>
>>>>>>>> jalemkul at vt.edu | (540) 231-3129
>>>>>>>> http://www.thelemkullab.com
>>>>>>>>
>>>>>>>> ==================================================
>>>>>>>>
>>>>>>>> --
>>>>>>>> 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.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>> ==================================================
>>>>>>
>>>>>> Justin A. Lemkul, Ph.D.
>>>>>> Assistant Professor
>>>>>> Virginia Tech Department of Biochemistry
>>>>>>
>>>>>> 303 Engel Hall
>>>>>> 340 West Campus Dr.
>>>>>> Blacksburg, VA 24061
>>>>>>
>>>>>> jalemkul at vt.edu | (540) 231-3129
>>>>>> http://www.thelemkullab.com
>>>>>>
>>>>>> ==================================================
>>>>>>
>>>>>> --
>>>>>> 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.
>>>>>>
>>>>>>
>>>>>> --
>>>> ==================================================
>>>>
>>>> Justin A. Lemkul, Ph.D.
>>>> Assistant Professor
>>>> Virginia Tech Department of Biochemistry
>>>>
>>>> 303 Engel Hall
>>>> 340 West Campus Dr.
>>>> Blacksburg, VA 24061
>>>>
>>>> jalemkul at vt.edu | (540) 231-3129
>>>> http://www.thelemkullab.com
>>>>
>>>> ==================================================
>>>>
>>>> --
>>>> 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.
>>>>
>>>>
>> --
>> ==================================================
>>
>> Justin A. Lemkul, Ph.D.
>> Assistant Professor
>> Virginia Tech Department of Biochemistry
>>
>> 303 Engel Hall
>> 340 West Campus Dr.
>> Blacksburg, VA 24061
>>
>> jalemkul at vt.edu | (540) 231-3129
>> http://www.thelemkullab.com
>>
>> ==================================================
>>
>> --
>> 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.
>>

-- 
==================================================

Justin A. Lemkul, Ph.D.
Assistant Professor
Virginia Tech Department of Biochemistry

303 Engel Hall
340 West Campus Dr.
Blacksburg, VA 24061

jalemkul at vt.edu | (540) 231-3129
http://www.thelemkullab.com

==================================================