[gmx-users] different springs - WHAM

[Justin Lemkul]

On 2/13/13 5:23 PM, Steven Neumann wrote:
> On Tue, Feb 12, 2013 at 5:01 PM, Justin Lemkul <jalemkul at vt.edu> wrote:
>>
>>
>> On 2/12/13 9:57 AM, Steven Neumann wrote:
>>>
>>> On Tue, Feb 12, 2013 at 2:53 PM, Justin Lemkul <jalemkul at vt.edu> wrote:
>>>>
>>>>
>>>>
>>>> On 2/12/13 9:40 AM, Steven Neumann wrote:
>>>>>
>>>>>
>>>>> Dear Gmx Users,
>>>>>
>>>>> I know it is possible to combine windows with different spring
>>>>> constants into the one PMF curve using g_wham.
>>>>>
>>>>> Do I have to somehow tell g_wham that one or two windows have
>>>>> different spring constants?
>>>>>
>>>>
>>>> No, they are read from the .tpr files.
>>>>
>>>>
>>>>> For instance - I got the better histogram overlap with lower force
>>>>> constant in one window. When I replace this window into the window
>>>>> with the sring constant like all windwos (worse overlap) both PMF
>>>>> curves differ app. 2kcal/mol which is around 30% of the overall
>>>>> deltaG.
>>>>>
>>>>> Is there any error I should inroduce when one window differ in terms of
>>>>> k1?
>>>>>
>>>>
>>>> What does g_wham's error analysis suggest?
>>>>
>>>> -Justin
>>>
>>>
>>> In both PMF error estimate with bayesian bootstraping is app. 0.2 kcal/mol
>>>
>>
>> Seems like a good result, so what's the problem?
>>
>> -Justin
>
> That the better overlap of histograms produce worse deltaG comparing
> to experiment. With all the same spring constants I get the
> experimental value of deltaG but there is a poor overlap. There must
> be (somehow) a correction added to deltaG when introdcuing windows
> with different spring constants.
>

The code and/or g_wham paper should address this.  In principle, WHAM can be 
conducted with any assortment of spring constants you like.  If you track down a 
bug or something, please report it on redmine.gromacs.org.

A workaround of course would be to simply add another window (or windows) with 
the same original force constant that gives adequate sampling.

-Justin

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

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

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