AMBER: Keq computations

From: Sean Rathlef <>
Date: Mon, 07 May 2007 15:10:09 -0800

Dear Amberites,

I have been trying to reproduce a known Keq value using an MMPBSA method in implicit water. My reaction is a host-guest (receptor-ligand) complex in the form of:

    R + L <---> RL (Hence, the Keq will have units of M-1).

I modeled the 'electrostatic energy in the solvated state' via MMPBSA for each species, and computed the gelG from these free energy values as follows:

    delG = G(products) - G(reactants) = G_RL - (G_R + G_L)

I then computed Keq via:

    Keq = e^(-delG/RT)

The expected Keq was 7.2, and my value was about 68,000. Seeing as this is a standard procedure, this value seems way too high. All species were modeled at the standard state of 1M.

Seeing as Keq is not unitless in this case, and in considering that the bimolecular step is dependent upon the concentration of L (ligand), where did I go wrong? Assuming that Keq is a *constant*, if we were to model the expected Keq (7.2) for this reaction (beta-cyclodextrin as host and 1-butanol as guest), would we need to model the relative energy for 1-butanol as a concentration dependent variable?

I guess the main question here is, how can I compute Keq and get a respectable result (i.e., ~7 M-1)? I'm not using Amber for this computation, but I would expect that parameterization of partial charges is somewhat conserved from package to package.

Any thoughts would be most appreciated,

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed May 09 2007 - 06:07:20 PDT
Custom Search