[gmx-users] Re: FEP softcore problematics

[mernst at tricity.wsu.edu]

After previous discussions that suggested separating vdw and charge mutations, I thought
it was literally a 2 stage process. In one stage I modify all charges (including those
for disappearing atoms going to zero) from their starting values to their final values,
using 10 lambda points and no soft cores. In another stage I modify vdw parameters from
their A states to their B states with soft cores, using only the final charge values for
A and B states, over 21 lambda points. My statistical error is enormously reduced from
when I was trying to modify charges and vdw simultaneously, but the results I'm getting
are still a bit puzzling.

It sounds as if I should actually be running 3 series of calculations -- is this
correct? I can:

1) Neutralize any atoms which are being disappeared
2) Modify LJ interactions from A state to B state
3) Change charges to final charges for B state.

or I can:

1) make the charges of the atoms present in the A-state go to those of the B-state
2) change the LJ of the A-state to the B-state.
3) make the charges of the atoms present in the B-state go to those of the B-state.

In the first series of calculations, there is one series of charge mutations that simply
changes the charges to 0 on disappearing atoms. The LJ interactions are changed while
atoms that will disappear are already set to 0 charge while atoms that do not disappear
continue to hold their starting charges. In the final series of calculations, atoms that
did not disappear in the B state are changed toward their final charges while the ending
state LJ parameters are used. Is this correct? I want to make sure I am not
misinterpreting anything.

In the second, equivalent series of calculations, the first job is to make all A-state
charges go to B-state charges. This would include both atoms that disappear and those
that do not disappear but that change parameters. The second job is to change the LJ
parameters from A-state to B-state while using... the B-state charges, I presume? I
thought that disappearing atoms shouldn't be charged as their LJ parameters change.
Under this interpretation, I'm not sure what the 3rd part means. I thought B-state atoms
already underwent mutation to B-state charges.

Matt Ernst
Washington State University