[gmx-users] BAR gives different result than TI

[Michael Brunsteiner]

Hi David,


i saw your recent post in the gmx mailing list about "BAR gives different result than TI" 
i wonder did you get any good answers to this queston so far!?

i was also trying to do some free energy calculations, with BAR i seem to get the correct
answer (compared to expt and literature) but there still seems to be some issues -
see: http://lists.gromacs.org/pipermail/gmx-users/2012-June/072867.html


also, i assume you used soft core potentials ... i remember quite a while ago i compared
results for free energies of solvation using  either sc_power=1 or 2 ... and the results
of (as i am sure, converged) calculations differed significantly. but this was a while ago
and in the meantime things might have changed.

another point: you write that your "numbers seem to have converged pretty good" ...

the molecules you use are obviously flexible and i recall talking once to a this guy
from some software company, they had done extensive FE calculations and he told
me that their main problem was getting converged results for the gas-phase leg of
the simulations, in your mail you only gave the net-results for the two molecules,
how do the two separate calculations (mutation in vacuum and mutation in solvent)
compare?

cheers
michael


===============================
Why be happy when you could be normal?