[gmx-users] convergence of PMF

[Justin Lemkul]

On 11/17/17 8:47 AM, Peter Kroon wrote:
>
> On 17-11-17 14:38, Justin Lemkul wrote:
>>
>> On 11/17/17 6:28 AM, abhisek Mondal wrote:
>>> Hello,
>>>
>>> I understand the invariant trend is the key feature to look out for
>>> in this
>>> case.
>>>
>>> But, is it possible to comment on the convergence from Force-vs-Time
>>> plot ?
>>> I mean, if I comment that using 0.01 nm ps-1 pull rate after 500 ps of
>>> pulling the force became zero and thus the biasing potential goes
>>> also to
>>> zero. Hence, after 5 nm of dissociation the PMF should attain
>>> convergence
>>> if also invariant trend-line is found near that point.
>>>
>>> Does this explanation make any sense from convergence of PMF's
>>> perspective ?
>> No, because you're conducting a non-equilibrium process, which is not
>> going to give you a PMF.
>>
>> Your system is changing over time due to the application of a biasing
>> potential. It does not make sense to say that anything there converged
>> based on any time-dependent quantity.
> But shouldn't you be able to look at the force-vs-time profiles of your
> windows? So that you can at least say that the force per window is
> equilibrated?

The OP mentioned a PMF, which is equilibrium sampling and indeed you 
*could* look at those forces, but then in this post it is mentioned that 
it's a non-equilibrium SMD simulation (non-zero pull rate), so no, 
there's nothing useful to look at here.

I would also just bypass the forces anyway, since they are extremely 
noisy, and compute the PMF over different windows of time. That's the 
most straightforward way to look at convergence (and what reviewers 
*always* ask to see in papers).

-Justin

> Peter
>
>> -Justin
>>
>>>
>>> On Thu, Nov 16, 2017 at 1:15 PM, Smith, Micholas D. <smithmd at ornl.gov>
>>> wrote:
>>>
>>>> One way to get some degree of confidence is to use the
>>>> bootstrapping/error
>>>> analysis options provided in gmx wham. But Justin is right you are
>>>> looking
>>>> for time invariance. So: use the first portion of your trajectory (say
>>>> first 60%) and compute the PMF, then use the first 65% of your
>>>> trajectory
>>>> and see if it has changed (and by how much), then 70%, and 80% and
>>>> so on.
>>>> If you have converged (within reason), you should see that at some
>>>> point,
>>>> as you add more time to the analysis the PMF doesn't change (in any
>>>> statistically significant way), when that occurs you have "converged".
>>>>
>>>>
>>>> ===================
>>>> Micholas Dean Smith, PhD.
>>>> Post-doctoral Research Associate
>>>> University of Tennessee/Oak Ridge National Laboratory
>>>> Center for Molecular Biophysics
>>>>
>>>> ________________________________________
>>>> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <
>>>> gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of Justin
>>>> Lemkul <jalemkul at vt.edu>
>>>> Sent: Thursday, November 16, 2017 8:04 AM
>>>> To: gmx-users at gromacs.org
>>>> Subject: Re: [gmx-users] convergence of PMF
>>>>
>>>> On 11/16/17 3:34 AM, abhisek Mondal wrote:
>>>>> Hi,
>>>>>
>>>>>        I have derived a PMF (protein-ligand) using umbrella sampling
>>>> method.
>>>>> But how can I comment if the obtained PMF has converged well ? What
>>>>> are
>>>> the
>>>>> ways to check the convergence of a PMF ?
>>>> How would you check convergence of any quantity? You have to determine
>>>> if the result is changing over time, and if it is invariant (within the
>>>> context of available statistics), then the quantity is converged.
>>>>
>>>> -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.biochem.vt.edu/people/faculty/JustinLemkul.html
>>>>
>>>> ==================================================
>>>>
>>>> -- 
>>>> 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.
>>>>
>>>> -- 
>>>> 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.biochem.vt.edu/people/faculty/JustinLemkul.html

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