FES is dependent on Biasfactor in W-MetaD

[Faye Tian]

Dear plumed users,

I am using well-tempered metadynamics to simulate weak acid dissociation reaction. 

First, I perform W-Metad to deposite a bias potential and generate HILLS file.

Then, I use the HILLS file as a static bias to perform long MD.

Finally, I reweight the MD traj to get the FES.

But I found that when I used different bias factors, the resulting bias potentials were very different, and the simulated FES was also very different.

I'm confused about the difference, any advice would be greatly appreciated.

[Gareth Tribello]

Hello

The free energy surface that you get from a metadynamics simulation is an estimate.  It can be different in different simulations because of random error.  I would recommend calculating the error bars on your free energy estimates.  You can find advice on how to calculate the errors in this tutorial:

I hope this helps

Gareth

On 2 Nov 2022, at 08:46, Faye Tian <feifei....@gmail.com> wrote:

Dear plumed users,

I am using well-tempered metadynamics to simulate weak acid dissociation reaction. 

First, I perform W-Metad to deposite a bias potential and generate HILLS file.

Then, I use the HILLS file as a static bias to perform long MD.

Finally, I reweight the MD traj to get the FES.

But I found that when I used different bias factors, the resulting bias potentials were very different, and the simulated FES was also very different.

I'm confused about the difference, any advice would be greatly appreciated.

<dist_vp_o_1_2_fes.compare.png>

<dist_vp_o_1_2_fes.compare.png>

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/136026ee-4445-4e18-9945-6abc35047e2dn%40googlegroups.com.
<dist_vp_o_1_2_fes.compare.png><dist_vp_o_1_2_fes.compare.png>

[Giovanni Bussi]

In addition to what Gareth wrote, consider that two profiles that you are only computing relative FEs, defined up to an arbitrary constant.

What I see from your FES plot is:

- FES at d>2 is visually very similar in the two runs (though shifted)

- FES at d<2 is visually very similar in the two runs

- The difference between the two regions has a difference (between the two runs) of approx 10 kJ/mol

My guess is that in the long MD at static bias you are seeing very few transitions between the two regions (low and high d). If this is what you see, then this is the explanation for the large statistical error

Giovanni

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/EF7739F1-A0E7-402E-8D69-55EDD3C6DA50%40gmail.com.

[Faye Tian]

Thank you. Here is a plot of the COLVAR file for a long MD run. I also plot the free energy difference between the two states (the dissociation energy). I'm not sure if they would converge to the same result under different static bias potentials after even longer steps.

You received this message because you are subscribed to a topic in the Google Groups "PLUMED users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/plumed-users/JDcxkAgKtAA/unsubscribe.
To unsubscribe from this group and all its topics, send an email to plumed-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/CAPLm8Z%2Bw_iPwJCVCf35o88zunkMD5pJwrfGD6U6Mr8NKMqoucA%40mail.gmail.com.

[Giovanni Bussi]

Hi,

your trajectory seems to have many transitions. It almost looks discontinuous... Are you sure there is no problem with PBC?

In theory, even with two different bias potentials, you should recover the correct populations.

Giovanni

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/CAMLiB%3D_x9dN2ODsDZxUBqBbpP74m8625gA65cp%2BWSgvnA3ghgA%40mail.gmail.com.

[Faye Tian]

The discontinuity comes from:
   - Output frequency is 5fs
   - CV is used to track proton, which is rapidly transferred between water.

The proton tracker code is from https://doi.org/10.1063/1.5027392.
Regarding PBC, I have attached my input file, which I think is ok.

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/CAPLm8Z%2BzOW9u2ODTH5oGaUoGh4QNWd2pi9vD9a97PpP-eSOAUQ%40mail.gmail.com.