AMBER Archive (2009)
Subject: Re: [AMBER] vlimit in TI with soft core potential
Date: Wed Jun 17 2009 - 03:47:38 CDT
> BTN, recharging biotin. But the free energy changes in the first and the
> third steps are very large. But we only care about their difference,
> which is relatively small. I think this process will bring large errors
> into the final result. So I want to try a 2-step way, in which the
> charges of IMI change to those of BTN and then the vdw parameters. The
The 1st and 3rd step dvdl can be very large if e.g. 1-4 interactions are
removed. That does not necessarily mean the corresponding result is
inaccurate. The electrostatic results are normally well converged,
especially the part stemming from 1-4-EE and the large free energy changes
should cancel out well.
If you have charges left on some softcore atoms in step 2 of your 2-step
process, the simulation will probably crash at some lambda point, as soon
as the vdw is softened enough for a solvent atom to crash into the exposed
charge. That is the reason why the 3-step process is necessary for now. We
are working on a better way to do TI in less steps, but it will have to
wait for Amber11...
Dr. Thomas Steinbrecher
610 Taylor Rd.
Piscataway, NJ 08854
AMBER mailing list