Re: Problems deciphering fepout

From: Floris Buelens (floris_buelens_at_yahoo.com)
Date: Sun Jan 18 2009 - 06:13:34 CST

Hi Chris,

If I've understood correctly what you're trying to do, it sounds it'll be next to impossible to get meaningful results from this approach... perturbing out an entire protein (or half a dimer) is simply too big a transformation to be able to quantify the free energy change. In theory the method can handle any perturbation but in practice anything remotely protein-sized will require a vast amount of MD time to get a converged result. I would suggest you start out with something like a solvation free energy calculation for a small molecule (e.g. fade its interactions in solvent surroundings), which will give you a feel for the amount of MD time required to get a converged free energy estimate, and move on from there. 15000 steps for each lambda value is unlikely to be enough even for a small molecule.
Best wishes,

Floris

----- Original Message ----
From: Christopher Hartshorn <cmhartshorn_at_wsu.edu>
To: namd-l_at_ks.uiuc.edu
Sent: Friday, 16 January, 2009 22:28:21
Subject: namd-l: Problems deciphering fepout

Hello. I am having trouble deciphering the output from the fep calculations of NAMD. Briefly, I am performing the calculation to measure dimerization of two proteins. Where I have two separate systems already equilibrated, etc. One of which contains the dimerized protein and the other sim has the two monomers separated in space. Both in the same solvent system, etc. I have chosen to fade one 1/2 of the dimerized protein out (columnB=-1) in the one sim and fade one of the monomers out (column B=-1) in the other sim. I am using the fep parameters from below and I am not sure of my out file's values.

1) Do the dG's for each step always change w.r.t. the Lambda step? As an example, one step in my calculation (0.3-->0.4) starts with -15KCal/mol and ends (after 15000 steps) with -30KCal/mol this is the case for all other Lambda steps. I ask because I thought that these values would not change much (e.g. would vary slightly about some constant value) because they were direct measurements done thousands of times over to get a good sampling.

2) What is plotted when I look at the tutorials on FEP? I see that it is dG vs. Lambda (0-->1), but where are the dG's coming from for each Lambda step in this graph? Is it the -62 or -29 value from my example here:
(#Free energy change for lambda window [ 0.4 0.5 ] is -29.3926 ; net change until now is -62.3803)?

source fep.tcl
fep on
fepFile dimer.fep
fepCol B
fepOutFile dimer.fepout
fepOutFreq 10
# LOOP OVER LAMBDA-STATES
fepEquilSteps 1500
set nSteps 15000
set dLambda 0.1
set init {0.0 0.000001 0.00001 0.0001 0.001 0.01 0.05 0.1}
set end {0.9 0.95 0.99 0.999 0.9999 0.99999 0.999999 1.0}
runFEPlist $init $nSteps
runFEP 0.1 0.9 $dLambda $nSteps
runFEPlist $end $nSteps

Thank you for any time you have to help.

Chris Hartshorn
WSU

      

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:52:15 CST