[gmx-users] general questions on re-run

[MD]

Hi Gromacs folks,

I am trying to look at protein-ligand interactions by running it multiple
times with different velocities. I also wonder if it is reasonable to
manually change initial coordinates of the ligand to use as a starting
point too, e.g. manipulating the ligand by rotating and translating it
slightly but not too far from the original position?

Or, do you use the first simulation result to start a new run, e.g. use the
largest cluster conformation as initial structure for another round of run?
Or do you use both?

How many runs do you usually do before analyze the whole set of data for
convergence?

Thanks in advance.

Ming