I performed a 50 ns production simulation for a protein in an organic solvents using NAMD. Now I want tp run a repetition for the simulation. I read about different approaches but I confused. can you guide me for the best way?
You want to repeat or extend the simulation? If you want to repeat then simply take the starting system and do simulation. If you want to extend then take the last frame of your 50ns simulation and start simulation keeping other parameters sane and at the end concatenate two trajectories.
You want to repeat or extend the simulation? If you want to repeat then simply take the starting system and do simulation. If you want to extend then take the last frame of your 50ns simulation and start simulation keeping other parameters sane and at the end concatenate two trajectories.
To repeat a simulation, you need use the same random seed. I suppose you didn't set random seed for the first run. In that case, the program will generate one base on your system time. You can find it from the output file. In the second run, use the same seed by setting the seed parameter explicitly to the one you find from 1st run's output file.
http://www.ks.uiuc.edu/Research/namd/2.6/olddocs/ug/node26.html
If you want to repeat the simulation, i.e., to obtain exactly the same trajectory, use the same seed for generating the initial velocities (viz. the answer by Peng Lian).
If you want to run another molecular dynamics simulation for the same system, for the purpose of obtaining more data, which is certainly a good idea, simply change the seed for generating the initial velocities and you will obtain different trajectory.
Dear Martin Klvana, what I want to run another MD simulation to prove that my data is reproducible. As far as I understood is changing the seed number will generate different trajectory and since I will keep the other conditions the same I have to get the same data. Am I right ?
Dear Peng Lian,
If I keep the seed number the same it seems that there is no meaning for repeating the simulation or I don't understand the idea of repeating clearly?
Mohamed: If you accidentally delete your trajectory data, you can recalculate the simulation using the same seed---and you will obtain exactly the same trajectory. If you write a paper about your simulations, by providing the initial structures and input files (including the seed numbers) you will make it possible for others to recalculate the simulations.
Different seeds will result in different initial conditions, namely different initial velocities, and therefore different trajectories of the system. This is a strategy used for enhancing the configurational sampling of the system and for assessing the uncertainty in obtained energies. If you run only one MD simulation you can only assess the convergence, e.g., by comparing average energies from the first and second half of the trajectory, but this is inferior to the unbiased comparison of parallel trajectories.
- Badges
- Science topic