[gmx-users] Re: free energy tutorial

[mernst at tricity.wsu.edu]

Hi Bharat,

This is the same tutorial that led me astray when I first began trying free energy
calculations in Gromacs. I am not sure how close to the experimental value the author
intended to get with this hydration free energy tutorial. If you are using a recent
version of Gromacs, be aware that the soft core method has changed. The tutorial says
sc-alpha should be 1.51, but that is more appropriate for the older method corresponding
to sc-power=2 rather than the sc-power=1 that is now recommended.

This tutorial also has you simultaneously mutate partial charges and atom types. I have
recently been informed, and confirmed via observation, that performing the partial
charge mutation with soft core potentials on can lead to misleading and very noisy
results. Look in the recent mailing list archives for suggestions and papers about
splitting the transformation into two series of calculations, one with partial charge
mutation and no soft core potentials and the other with soft core potentials and no
partial charge mutation. Even if you don't want to do this for the simple tutorial
you'll probably want to do it for real systems in the future.

Finally, the tutorial gives the dubious advice to use the final structure from one
window as the starting structure for the next window. If you instead run each window's
simulations completely independently, you can simultaneously run calculations on several
machines and don't have to worry about the output of previous windows somehow biasing or
contaminating the current window. Again, I don't know if it is necessary to operate this
way to get good results on this simple tutorial, but it does become more important if
you want to run novel, non-trivial systems in the future.

Matt Ernst
Washington State University

> Dear GMX users,
>
> i am trying 'relative hydration free energy of p-cresol wrt toluene' tutorial
> from http://md.chem.rug.nl/education/Free-Energy_Course/3.mutation-fe.html. i am using
> all the mdp files given on
> the site, the confout.gro from toluene/eq directory of the previous tutorial as starting
> structure for this simulation.
> and whatever relevant changes has to be made in top and gro have been made. then i
> performed the simulation for lamda
> values from zero to one with an interval of 0.4, so total of 26 simulations. but i am
> not getting the mentioned
> experimental value of ddG-hyd.
>
> i got the following values in rmsd.xvg
>
>         IN WATER                    IN VACUO
> ==========================   ============================
> 0    -61.874991   2.59103    0    -98.521845   8.80653
> .04  -57.180709   1.73975    .04  -67.562344   0.237903
> .08  -58.786951   0.266783   .08  -64.780817   1.23656
> .12  -60.764221   0.300319   .12  -63.280561   1.14712
> .16  -62.700774   0.313339   .16  -60.123999   0.926409
> .20  -65.107642   1.26388    .20  -58.584394   0.103875
> .24  -63.749578   0.384702   .24  -56.668703   1.02234
> .28  -64.267572   1.22184    .28  -54.837984   0.0807904
> .32  -61.524424   0.402878   .32  -53.550764   0.0760657
> .36  -55.897929   0.314078   .36  -51.510805   0.0622261
> .40  -49.084890   1.05137    .40  -49.736461   0.0627338
> .44  -38.339801   1.39297    .44  -48.470839   0.0718284
> .48  -29.613058   1.38755    .48  -46.842750   0.0740531
> .52  -24.891702   1.04125    .52  -45.007666   0.0766915
> .56  -25.435188   1.70549    .56  -43.939682   0.101024
> .60  -25.180946   1.82089    .60  -41.844138   0.111544
> .64  -28.395296   0.849711   .64  -41.250909   0.13644
> .68  -29.672103   2.26016    .68  -41.182423   0.139366
> .72  -34.441109   2.00941    .72  -41.747919   0.127424
> .76  -39.148106   2.01051    .76  -44.426396   0.143008
> .80  -49.360365   2.34239    .80  -44.381975   0.135462
> .84  -57.031557   2.79298    .84  -45.564523   0.123685
> .88  -64.059494   2.79089    .88  -46.074452   0.104784
> .92  -78.668957   1.97957    .92  -46.678656   0.0993529
> .96  -85.671014   1.92346    .96  -46.871521   0.0923201
> 1.0  -52.956186   1.26136    1.0  -47.183714   0.0934921
>
>
> upon integration in xmgr, following are the dG values:
> dG-wat = -50.6556 kJ.mol-1
> dG-vac = -51.1109 kJ.mol-1
>
> ddG-hyd = 0.455 kJ.mol-1
>
> the experimental value mentioned on the site is -22.1 kJ.mol-1
>
> Please help...
>
> bharat