AMBER Archive (2002)

Subject: expression of V(lambda) in GIBBS

From: hagop demirdjian (hagop.demirdjian_at_ens.fr)
Date: Tue Sep 24 2002 - 07:44:02 CDT

Hi AMBER list,

here are a few questions I have regarding Free Energy Perturbation in
AMBER :

What is the expression of V(lambda) in GIBBS (I could not find it in the
manual, so I assumed it is linear : V(l) = l*V1 + (1-l)*V0

I simulated a mutation where atoms disapear. I computed DeltaGs in GIBBS
and SANDER :
in GIBBS, dV/dl values are all negative and decrease from -16 to -30
kCal/mol/l
in SANDER, I used klambda = 4, and got values decreasing quickly from
350 to -20 and increasing to 0 at lambda =1.

Both integrations give the same DeltaGs. I assume that differences are
due to a difference in the expressions of the V(l). Am I right ?

I also would like to know what IDSX0 parameter people usually use in
GIBBS to account for vanishing atoms. Is 500 a good value ?

And finally, what in your opinion is best suited for TI with vanishing
atoms ? : SANDER with klambda = 4 or GIBBS with an appropriate value ISDX0 ?

Thank you for your answers.

Hagop. 

-- 
Hagop Demirdjian
Département de Chimie
Ecole normale supérieure
24, rue Lhomond
F-75231 Paris Cedex 05

Tél. 01 44 32 24 10