GROMACS version: 2023.3
GROMACS modification: No
System is approximately 150000 atoms (protein-protein complex) and was parametrized in Charmm36 force field. I have done NVT and NPT equilibration for 1 ns each (V-rescale termostat and C-rescale barostat). The target pressure value was 1 bar. After NPT the average pressure value was -1.0 ± 100.
Of course, pressure may fluctuate, but is the obtained result acceptable for the further molecular dynamics?
It is difficult to say what is acceptable. You probably know your system better. It also depends on the temperature. Check your pressure (or box volume) fluctuations, as well as, e.g. the potential energy of the system. Is it stable in the last half of your NPT equilibration? I would expect your equilibrations to be a bit too short.
I would suggest to run at least 10-20 ns NVT equilibration and 20-100 ns NPT equilibration. What is important is that the system is stable in the last half (or quarter) of the equilibration.
Thank you very much for your advice
Increasing the time of equilibration actually allowed me to bring the pressure value closer to the target of 1 bar.
At the same time, long simulations require more disk space.
Is it permissible to save coordinates, energy, etc at a lower frequency (via the nstxout-compressed, nstenergy etc parameters in the .mdp file)? Changing the parameters from 500 to 5000 would make the files smaller, especially the .xtc one.
Won’t this affect the accuracy of further molecular dynamics simulations?
I think writing output too frequently is a common mistake in our field in general. The number of output frames should be adapted to your needs - never just base it on examples, or what others in your group are using or have been using. For equilibrations I would recommend a total of 5 to 50 coordinate frames (nstxout-compressed often in the range of 10,000 to 5,000,000) depending on if I expect that I will need to visually inspect them. For production runs you might want more frames, especially if you want to carefully analyse what is happening and/or render a movie, but even so I think nstxout-compressed in a range of 10,000 to 1,000,000 is reasonable. You should have specific reasons for writing nstxout, nstfout and nstvout at all - if you need them you probably know why you need them. Energy output (nstenergy) only needs to be written as often as you need for your analyses, i.e. to follow the evolution of the observables you are interested in and to make the plots you need. For equilibration you do not need to output energy very often at all.
Regarding accuracy of follow-up simulations, it is not at all affected by the output intervals. When starting your next simulation you can start from the confout.gro from the simulation before.