[gmx-users] RE: Free energy protocol

bmbbl at leeds.ac.uk bmbbl at leeds.ac.uk
Wed Jun 15 15:48:01 CEST 2005

Thanks Groetjes.
I also have a couple of detail questions.

1) I should swtich off -DPOSRE option during step 1 and 3, right?
2) how long, do you think, is suitable to the rerun mdrun in step 3? Can I use
the same period of time as step 1, e.g. 1.25-2.5 ns?

Thanks a lot.

> Hi there,
> Since nobody has replied my post, I am sending it around once more.
> I want to confirm my protocol of free energy difference calculation using FEP
> with GROMACS.
> A-B and A-B' complex
> 1. Full MD on A-B complex. input 1.tpr; output: 1.xtc, 1.gro
> 2. perturbation topology indicated as: B --> B' names as newtp.itp
>     grompp -f FEP.mdp -o 2.tpr -c 1.gro -p newtp.itp
> 3. mdrun -c 1.tpr -rerun 1.xtc -o AB.edr
>    mdrun -c 2.tpr -rerun 1.xtc -o AB'.edr
> 4. g_energy -f AB.edr -f2 AB'.edr -fee -o FE.xvg
> My question is on step 2 and 3. Do I have to run full MD using 2.tpr and 1.gro
> to generate a new 2.xtc file or I can simply apply 1.xtc to the -rerun?

No. The Free Energy Perturbation (FEP) assumes that the perturbation
does not (significantly) affect the enemble, only the energies. So, you
must in both cases (2 and 3) use the same trajectory (1.xtc).

However, as an alternate check, you can apply the perturbation the other
way around as well, and use mdrun -c 2.tpr to generate '2.xtc', and then
rerun 'mdrun -rerun 2.xtc -c 1.tpr -o 2AB.edr' and 'mdrun -rerun 2.xtc
-c 2.tpr -o 2AB'.edr'. In principle, this must give the same answer, but
in practice you will see a (small?) difference, which is a good
indication of the actual error you make in your method.

By the way, I'm not sure if you have all the file options correct in
your examples above. Check the manuals of mdrun, grompp and g_energy.

Also, as a side remark, it may not be a good idea to actually use a
prime (') in a filename (as in AB'.edr in your example).

More information about the gromacs.org_gmx-users mailing list