Re: Judging Convergence on Free Energy Calculations

From: Josh Vermaas (
Date: Fri Jan 15 2021 - 14:13:13 CST

Hi Ethan,

What I like to do is analyze different snapshots of my simulation. If
the first half of the trajectory and the last half of the trajectory
yield the same PMF (within some tolerance) when analyzed independently,
I judge the PMF to be converged. Sometimes, the first few nanoseconds
aren't representative of the rest, and so it can also be useful to
analyze the PMFs from 5-10 nanosecond chunks of trajectory.


On 1/15/21 2:45 PM, Ethan Croitoru wrote:
> Hi,
> I am trying to run free energy calculations on a transmembrane helical
> dimer. I have been trying to closely follow the protein
> ligand tutorial to do these calculations
> (
> <;!!HXCxUKc!jr-kr8EkCJJ3A7k6RXBLqnfo4fRtuwckGfuofrJzRv2oa7zAUADT2qmVIUHnPcY$>).
> I have a question about judging convergence of the PMFs. Taking the
> bound conformation calculation as an example, using the RMSD colvar,
> could one look at the protein's rmsd plateauing as a sign that the
> PMFs are also converging. Or should I be looking for my PMF values to
> also be plateauing? In a similar vein, around how many steps should
> convergence be happening around.
> Thank you for the help,
> Ethan Croitoru

This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:10 CST