> First of all, thanks again for the contributions. I have one last question.
>
> To check for convergence, does it pay to run a relatively long simulation, say 60ns, or is it better to run 6 independent simulations of 10ns each?
>
> The reason I'm asking is because I have found that in protein-ligand complexes the starting conformation of the ligand may have an influence on measuring a particular property (deltaG in my case). I've started MD simulations from different docking poses and have found that the ligand may be unable to overcome low minima in the time frame of my simulations.
How do you prepare the protein-ligand complex? Do you have a XRAY
structure, or model it? Or do you prepare a protein solvated with
water+ligand and then try to see the binding? If there is a known
structure for the protein-ligand complex, then starting from that
conformation and doing a long simulation is a good approach. Multiple
simulations are always the best, so run a couple of such simulations.
Or if you start from scratch and try to see how a free ligand will bind to
the protein, you will need to run multiple simulations. Even that method
might not help, though. If the binding happens on the surface, then you
might see the binding, but what if (in reality) there is a big global
conformational change in the protein that will bind the ligand to its
inside? Again, giving an example from nucleic acids, loop regions in RNA
are dynamic and it is hard to find how the binding of a ligand will be on
that loop regions. It is very normal to see big global change in the loop
that will affect the ligand binding. The same thing might or might not
happen in your case, but keep this in mind.
Namely, you are right that if there are multiple minimas with big energy
barriers, you might not see the transitions in your MD time. It is always
better to run multiple simulations to scan the conformational space in an
MD simulation if the system is dynamic (but again that might not solve
the sampling issue). Replica exchange is another method.
Ilyas,
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Jul 07 2012 - 11:00:02 PDT
