[gmx-users] BAR histogram results

[Justin Lemkul]

On 2/1/13 6:07 PM, Sonia Aguilera wrote:
> Hi!
>
> I just finished a FE calculation using the BAR method. This is the plot of
> my histogram.xvg file:
>
> <http://gromacs.5086.n6.nabble.com/file/n5005095/Screenshot.png>
>
> It is my understanding that data must be overlaped in order to get adequate
> sampling (I have been following Justin Lemkul's tutorial
> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/free_energy/07_analysis.html).
>
> However I have a few questions (First two must be really simple, but I just
> have not been successful):
> 1. How can I plot the ∂H/∂λ values separately from the ΔH values? I have
> been trying, but I'm not really an expert using xmgrace.

You will have to parse the .xvg file, either manually or with a script, in order 
to separate the data sets.

> 2. I also tried to plot this data using gnuplot. However, this program does
> not distinguishes between diferent leyends, as in the xmgrace plot. Is there
> any way for plotting the this data using this program?

All Gromacs programs allow you to turn off XmGrace-specific formatting with 
"-xvg none."

> 3. Comparing my plot with the one on the tutorial, something is wrong with
> my sampling. I have been trying to understand Bennett's paper, but I have
> not been successful. Can someone explain to me the meaning of the values in
> each axis? How should I interpret this data? What can I do to improve my
> sampling?
>

In a very basic sense, the result is a series of histograms of energy values 
from neighboring lambda windows.  You need continuity in sampling from lambda = 
0 to lambda = 1 in order to establish a reliable value for deltaG.  It seems 
from your plot that you are only using a few lambda points, with large gaps 
between 0.1, 0.5, and 0.9.  You probably need more windows.

-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

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