Re: nonuniform sampling in ABF

From: Jérôme Hénin (
Date: Tue Jun 12 2018 - 17:09:56 CDT


Haohao's answer is very good. I would also add that the long-time behavior
is uniform sampling, but for a biological system even those timescales are
very hard to achieve, so the convergence is local anyway. A way I like to
look at the sampling histograms is to calculate the associated effective
PMF, that is, 0.6*ln(N(x)), in kcal/mol at 300 K. Then I compare this
effective PMF to the calculated PMF from ABF. this tells me by how much the
free energy landscape has been flattened.


On 12 June 2018 at 18:25, Haohao Fu <> wrote:

> The problem is that the reaction coordinate model chosen is too simple to
> describe the process. Your definition of RC implies that the rotational
> movement and the conformational change of the proteins can be observed in
> the time scale of your simulation, which is not true.
> Just try to introduce some restraints to reduce the degrees of freedom
> (e.g. Orientation、polarPhi and polarTheta colvar) and discuss the effects
> of the restraints. See J. Chem. Theory Comput., 2017, 13 (11), 5173–5178
> and J. Chem. Theory Comput., 2013, 9 (1), 794–802 for some examples

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:21:10 CST