[gmx-users] Different TI free energy values in 4.0.7 and 4.5.3

[chris.neale at utoronto.ca]

Based on your use of:

sc-sigma                 = 0.3

I wonder what happens with 4.5.3 when you set the env.var.  
GMX_SCSIGMA_MIN to 0

quoting http://www.gromacs.org/About_Gromacs/Release_Notes/Versions_4.5.x

"for free-energy calculations sc-sigma now also sets the minimum  
soft-core sigma (old tpr files retain the old behavior, which can be  
enforced by setting the env.var. GMX_SCSIGMA_MIN to 0)"

-- original message --

Dear Gromacs users,

    One example of my mdp files with lambda=0.5 is like the following:

integrator               = md
tinit                    = 0
dt                       = 0.004
nsteps                   = 250000
simulation_part          = 1
comm_mode                = Linear
nstcomm                  = 10


nstlist                  = 2
ns_type                  = grid
pbc                      = xyz
periodic_molecules       = no
rlist                    = 1.4


coulombtype              = PME
rcoulomb-switch          = 0
rcoulomb                 = 1.4
epsilon_r                = 1
epsilon_rf               = 1
vdw_type                 = cut-off
rvdw                     = 1.4
fourierspacing           = 0.168
pmeorder                 = 4
ewald_rtol               = 1.0e-5
ewald_geometry           = 3d
optimize_fft             = yes


tcoupl                   = v-rescale
tc-grps                  = ioq !ioq
tau_t                    = 0.1 0.1
ref_t                    = 298.15 298.15
pcoupl                   = Berendsen
Pcoupltype               = Isotropic
tau_p                    = 1.0
compressibility          = 4.5e-5
ref_p                    = 1.0


constraints              = all-bonds
constraint_algorithm     = lincs
lincs_order              = 4
lincs_iter               = 1

free_energy              = yes
init_lambda              = 0.5
delta_lambda             = 0
foreign_lambda           =
sc-alpha                 = 0.7
sc-power                 = 1
sc-sigma                 = 0.3
nstdhdl                  = 10
separate-dhdl-file       = yes
dhdl-derivatives         = yes
dh_hist_size             = 0
dh_hist_spacing          = 0.1
couple-moltype           =
couple-lambda0           = vdw-q
couple-lambda1           = vdw-q
couple-intramol          = no

     The system has 1 Chloride ion with ffG43a1 force field parameter and SPC
water molecules with heavy hydrogen in a cubic box of size 3 nm. I have
sampled 21 lambda windows ranging from 0 to 1.

    The values are converged by block averaging.

        Regards,
            Yan

On Thu, Apr 14, 2011 at 3:35 PM, <chris.neale at utoronto.ca> wrote:

> Dear Yan:
>
> If you want to get to the bottom of this then (a) please report your mdp
> files and (b) show some evidence that your values are converged by, for
> instance, block averaging the run-time.
>
> Chris.
>
> -- original message --
>
> Dear Gromacs users,
>
>  I got different free energy values by Thermodynamic Integration (TI) in
> Gromacs 4.0.7 and 4.5.3.
>
>  When I calculated the hydration free energy of Chloride ion by using
> Thermodynamic Integration method, the results for the free energy of
> un-charging the Chloride ion from q=-e to 0 in SPC water are 359+/-6 kJ/mol
> in Gromacs 4.0.7 and 385 +/-6 kJ/mol in Gromacs 4.5.3, respectively.
>
>  The literature value is 392 kJ/mol (J. Phys. Chem. 1996, 100, 1206-1215,
> Table 6), which is close to the result in Gromacs 4.5.3 but not in 4.0.7.
>
>  As a check, Gromacs 3.3.1 gave the result as 381+/-3 kJ/mol.
>
>  I searched the mail list and the bugzilla of Gromacs but didn't find
> explanation for this issue of version.
>
>  Does anyone know the reason for the different TI free energy values in
> those different version of Gromacs?
>
> --