AMBER Archive (2009)

Subject: Re: [AMBER] 2-step vs 3-step TI, and softcores question

Date: Tue Jun 23 2009 - 17:27:46 CDT


I am not sure that the two results should be equal at all, you are
changing different potential functions, after all... Are the overall
resulting deltadeltaGs compareable? Also, are you sure your simulations
are converged?

> So the electrostatics contributions calculated using different methods
> are different, unless I am mistaken and should subtract the charge
> addition contribution from the charge removal contribution in the
> 3-step method, which would give me approximately the same answer as
> the 2-step method. 23.84 ~ 24.33. Is this the correct solution to this
> discrepancy, or have I made a mistake somewhere in my calculations?

The results for removing charges, changing vdw and adding charges again
should be added together.

> Also, why is it that softcore potentials used for the vdW transition
> in the AMBER 10 3-step TI method cannot use clambda values <0.01 or
>>0.99? This prevents the use of 12-point gaussian quadrature because
> the 1st and 12th quadrature points fall outside of the allowed range.

This is related to the way softcore-atom forces are handled internally.
You could use linear interpolation for the very end of you dvdl-curve.

Kind Regards,


Dr. Thomas Steinbrecher
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854

AMBER mailing list