Re: FEP output in NAMD

From: Jerome Henin (
Date: Fri Apr 08 2005 - 13:37:41 CDT

> I have just started reading/understanding the free
> energy perturbation method implemented in NAMD. I have
> some questions:
> (1) Can these calculations be done under NPT
> conditions? or only NVT?
The basic theory is developed for NVT, but for most practical purposes,
switching to NPT does not make a significant difference. For that reason, the
code is routinely used in NPT.

> (2) I have a protein with two ligands L1, L2 and iam
> attempting to compute the relative free energy of
> binfing between these two ligands. For one of the
> ligand I have the crystallographic positions and for
> the other, i have taken the ligand coordinates after
> 5ns of MD simulation. if i did not turn on the
> "minimize" option in the configuration file, i had
> problems with RATTLE. So i turned it on for sometime
> and then FEP resumed. Iam not sure if this is OK.
I am not sure I get your point here.

> (3)From the manual, I understand that "dE_avg" is the
> accumulated energy ensemble average and "dG" is
> corresponding free energy at each step.
To be precise, dE is the ensemble average, in the initial state, of the energy
difference between the initial and final states.

> What is this
> "dG" that is calculated?
Using the FEP expression with the in-progress value of the ensemble average of
the exponential (see manual for the FEP expression)

> to calculate average energy
> for a particular window, the manual says "dE_avg"
> should be summed.
To compute the average energy *difference*, yes.

> is this what goes into the final
> equation to compute overall relative free energy of
> binding:
> sum(window_size*avergae_energy in that window)?
No. Again, see FEP expression in the manual.


This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:40:39 CST