Re: [AMBER] how to choose the final binding mode

From: Jason Swails <>
Date: Mon, 26 Jan 2015 10:27:20 -0500

> On Jan 26, 2015, at 6:23 AM, 徐晓庶 <> wrote:
> Hello Amber user:
> I'm finished the MD simulation now. I want to analyze the binding mode between the protein and a small ligand, I'm confused how to choose the pose. As there will be a lot of poses generated during MD simulation, should I choose the pose of the last moment?

Typically, no. That pose is arbitrary. I assume that since you’re asking this question, that an arbitrary choice is not ideal.

> Or do you have good suggestions about how to choose ?

Pick the “dominant” pose (i.e., the pose that exists in more frames than any others). You can also cluster the poses and pick the representative structure of the most populated cluster. Or you can histogram a few geometrical measures that you use to ‘define’ the pose (certain distances or angles, for example), and pick a structure that falls in the most probable (i.e., most populated histogram).

These are all ideas based on the assumption that you *need* to pick only a single pose, but you should make sure this is true. The true ensemble is built from all of the poses you see in your simulation (and more) sampled according to the Boltzmann distribution. So it really depends on what you want to do -- if you want a free energy, *no* single pose will suffice (you really need all of them).


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Mon Jan 26 2015 - 07:30:03 PST
Custom Search