FEP double annihilation

From: Carlo Guardiani (carlo.guardiani_at_dsf.unica.it)
Date: Fri Sep 23 2016 - 13:28:30 CDT

Dear NAMD experts,

I am trying to compute the free energy of binding of
a calcium ion to an ion channel and I would like to
use the double annihilation technique that in the FEP
tutorial was illustrated in the example of potassium
binding to the 18-crown-6 ether.

I have already performed a long NPT simulation of my
system that comprizes the channel embedded in the
membrane and solvated on both sides by a CaCl2
solution. I am trying to perform both annihilations
starting from the same conformation of the NPT
trajectory. In the first simulation the annihilation
involves a single calcium ion in the Selectivity Filter
of the channel while in the second simulation I annihilate
a randomly chosen calcium ion in the middle of the bulk.
In both simulations the annihilated ion is harmonically
restrained to its initial position. The FEP section of
my input file was:


source fep.tcl

alch on
alchType FEP
alchFile Last_Frame_renumb_TER.fep
alchCol B
alchOutFreq 10
alchOutFile forward.fepout

alchElecLambdaStart 0.1
alchVdwLambdaEnd 1.0
alchVdwShiftCoeff 5.0
alchdecouple on

alchEquilSteps 100
set numSteps 1000

set dLambda 0.01

runFEP 0.0 1.0 $dLambda $numSteps

In both simulations the behaviour of the free energy
as a function of lambda is quite strange. While in
the crown ether tutorial the free energy grows
linearly until it reaches a plateau, in my case
I have a change in slope. For lambda < 0.5 in each
lambda interval the free energy change is approximately
dG=5. For lambda slightly higher than 0.5 the free
energy becomes negative for a few lambda intervals.
After that the free energy grows again until it
stabilizes around dG=30. I am new to free energy
calculations so I don't know exactly what to expect.
Is this behaviour normal or should I get a linear
plot as in the crown ether example ? What are the
possible reasons of this behaviour and how can
I avoid it if it is not normal ? I checked the
trajectory of the simulation but I didn't find
anything strange.

Thank you very much for your help and kind regards,

Carlo Guardiani

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:22:29 CST