GROMACS version: 2020.3/2021.3
GROMACS modification: No
I’ve noticed Gromacs 2021.3 can read setup.tprs from Gromacs 2020.3. Is it not always better to use the same version of gromacs for simulation set up and then the final md run?
GROMACS version: 2020.3/2021.3
GROMACS modification: No
I’ve noticed Gromacs 2021.3 can read setup.tprs from Gromacs 2020.3. Is it not always better to use the same version of gromacs for simulation set up and then the final md run?
Hi,
If the .tpr
file can be read by the more recent version it does not matter. It means that the file format has not changed between the two versions (unless a bug has been introduced).
Now, since the file format does change every so often it is, in general, best practice to use the same version for set-up and run. But if it works it works. If it does not work, Gromacs will stop and report an error.
Petter
In some cases, the logic for automatically-filled values (or default values) can change between versions. If you use TPRs generated with older versions of grompp
, you should know that a newer gromacs installation might generate a slightly different TPR from the same (sparse) mdp file. In this case, the -po
output from grompp can be helpful.
Assuming no thermodynamically relevant bugs, the results for simulations run from the same TPR file should be thermodynamically equivalent across releases. If you are comparing trajectories produced form different versions, though, you should probably check the release notes for relevant changes to automatic behaviors or default values. More recent releases of grompp may also have improved safety checks (or performance-related advice) for the parameter choices in the mdp file.