Re: Problem of calculating PMF changed by tilt angle using extended ABF

From: Jérôme Hénin (
Date: Mon Jun 07 2021 - 05:30:26 CDT

Dear Namho,

> We wanted to determine free energy change upon rotation of a molecule using
> 'angle' colvar sampled by extended ABF

I don't think you need the extended-system version of ABF for this, standard ABF should work for an angle coordinate. That said, it will become necessary if you switch to more complicated coordinates, so you can keep it.

> So we tested our methodology on an only-water system which consist of 251 water
> molecules in 20 X 20 X 20 A sized box.

> In the system the center of mass of a selected water molecule was fixed at the
> center of the simulation box using harmonic constraint and was only allowed to
> rotate freely.

I don't know how will-defined the angle will be, since the center of mass is not really fixed, but will fluctuate around the center of the restraint, giving a noisy version of the angle.

> We were trying to calculate the free energy change as a function of angle
> between the selected water dipole and the z axis using eABF.

In that case, you could constrain the oxygen atom (a real, hard constraint) using the fixedAtoms feature of NAMD, and measure the angle of the vector between the oxygen and the two hydrogens. That would match the dipole vector precisely.

> It is obvious that the free energy change should be negligible with respect to
> the angle because the system is isotropic.

It's not that obvious to me. An angle coordinate can have a non-zero Jacobian term - generally the probability would be proportional to the sine of the angle. You can test this by running a simulation without the ABF bias and collecting the histogram of the angle. Is it uniform?


This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:11 CST