Fundamental question about Hamiltonian replica exchange (HREX, especially REST2)

[Yifan Jin]

I just start to learn how to perform the enhanced sampling with Hamiltonian replica exchange. The tutorial on the official website (https://www.plumed.org/doc-v2.7/user-doc/html/hrex.html) provides this sample command to start the simulation after the .tpr files for all the replicas are generated:

mpirun -np $nrep gmx_mpi mdrun_d -v -plumed plumed.dat -multi $nrep -replex 100 -nsteps 15000000 -hrex -dlb no

But the flag "-multi" has been eliminated in the recent versions of GROMACS, and only the option "-multidir" is available. I am not 100% sure how to use it for HREX. I guess that, for example, if I have five replicas, should I create five sub-directories d1, d2, d3, d4, and d5, and place the five .tpr files inside correspondingly? And in the command above, should I just replace "-multi $nrep" with "-multidir d1 d2 d3 d4 d5"?

Another question is how to generate the .tpr file for each replica at different temperatures. In a normal MD simulation, the basic process is minimization, equilibration, and final MD. I guess that the file "processed.top" should be created using "grompp -pp" after equilibration, is that correct? The five temperatures in the tutorial from the link above are 300, 405.36, 547.723, 740.083, and 1000K. Should the equilibration be performed at these different temperatures? Or should I run the equilibration just at 300K and apply the various temperatures (in terms of scale) when using the partial_tempering tool?

I am confused about the description on the website. Could anyone who has done the relevant simulations (especially REST2) kindly share the basic procedure (it would be awesome if you can provide the commands and input files as well)?

Thanks in advance

[Giovanni Bussi]

On Fri, Sep 17, 2021 at 4:15 PM Yifan Jin <jin89...@gmail.com> wrote:

I just start to learn how to perform the enhanced sampling with Hamiltonian replica exchange. The tutorial on the official website (https://www.plumed.org/doc-v2.7/user-doc/html/hrex.html) provides this sample command to start the simulation after the .tpr files for all the replicas are generated:

mpirun -np $nrep gmx_mpi mdrun_d -v -plumed plumed.dat -multi $nrep -replex 100 -nsteps 15000000 -hrex -dlb no

But the flag "-multi" has been eliminated in the recent versions of GROMACS, and only the option "-multidir" is available. I am not 100% sure how to use it for HREX. I guess that, for example, if I have five replicas, should I create five sub-directories d1, d2, d3, d4, and d5, and place the five .tpr files inside correspondingly? And in the command above, should I just replace "-multi $nrep" with "-multidir d1 d2 d3 d4 d5"?

Yes, this should be correct

 

Another question is how to generate the .tpr file for each replica at different temperatures. In a normal MD simulation, the basic process is minimization, equilibration, and final MD. I guess that the file "processed.top" should be created using "grompp -pp" after equilibration, is that correct? The five temperatures in the tutorial from the link above are 300, 405.36, 547.723, 740.083, and 1000K. Should the equilibration be performed at these different temperatures? Or should I run the equilibration just at 300K and apply the various temperatures (in terms of scale) when using the partial_tempering tool?

Processed.top does not contain coordinates but just the force field parameters, so it is irrelevant if you do it before or after equilibration. You should however do it with your final setup (I mean: after having added all water and ions).

Simulations are all performed at a nominal temperature of 300K. The temperature indicated there is the "temperature of the solute" (in some sense). Mathematically, the only thing that matters is the inverse of the ratio between that temperature and 300, that will tell you how much you are scaling the parameters.

I think the example there is exactly for a REST2 simulation. In particular, if you "mark" with underscore all the solute atoms you will have the same protocol described in the original REST2 paper, with the full solute scaled, perhaps with some irrelevant different in how bonds and angles are scaled (in our implementation they aren't)

Giovanni

 

I am confused about the description on the website. Could anyone who has done the relevant simulations (especially REST2) kindly share the basic procedure (it would be awesome if you can provide the commands and input files as well)?

Thanks in advance

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/846addce-3f1e-4243-9a5b-f1f8fb6b9511n%40googlegroups.com.

[Yifan Jin]

Thanks so much. That makes a lot more sense