Using westpa to simulate a reaction with about 35 kcal/mol activation energy

[moutos...@gmail.com]

Dear concerns,

We would like to use westpa to simulate a hydrogen transfer reaction in a QMMM system where the activation free energy from umbrella sampling is about 35 kcal/mol. Is it possible to overcome 35 kcal/mol activation free energy and get the PMF using westpa? 

Thanks.

Regards-

Moutoshi

[Daniel Zuckerman]

Moutoshi, I can make a few comments about this.

First, as you probably are aware, if the barrier is truly 35 kcal/mol, the reaction would not occur in nature.  This could imply something wrong with the model (force field, qm/mm setup) and/or sampling.  In my biased opinion, the most likely suspect is the sampling that went into your umbrella sampling study, but you should carefully consider all facets of the issue.

Could WE handle a very large barrier?  In principle, yes, but it will depend on whether you have a good progress coordinate.  If the barrier is very high, you would probably need something close to the true reaction coordinate for success.  On the other hand, you can always just try and see if WE/WESTPA gets stuck or whether it continues to make progress on the coordinate you choose.  It is possible that slower re-arrangements of the protein are required to see the true barrier/path but WE might be able to sample some of this because such motions presumably are correlated with reaction progress.

I will also mention that the standard challenge of WE simulation - that it can be difficult to quantify rate constants even if you achieve success in generating rare events.  This is discussed in some of our papers - e.g., Annual Reviews 2017 and recent work by Adhikari 2019, by Mostofian 2019, and by Copperman 2020.  You need to decide what your goals are.  If obtaining pathways only (not rates) will be a success, then you should try WE.

If this were my system, I would first conduct a 'post mortem' to see what went wrong with umbrella sampling (e.g., sample some windows much longer and see if the distribution changes in those windows).  Then I would try WE with a simple progress coordinate and see if it makes progress.  If that fails, I would consider examining a higher temperature or even a targeted simulation to get an idea for better progress coordinates.

If you are not aware of it, we offer a 'data club' where you can discuss your particular problem with members of the Chong and Zuckerman groups, without any implied collaboration - no strings attached.  Let me know if you're interested.

Good luck.

--Dan Zuckerman

---

From: westpa...@googlegroups.com <westpa...@googlegroups.com> on behalf of moutos...@gmail.com <moutos...@gmail.com>
Sent: Wednesday, October 28, 2020 6:41 PM
To: westpa-users
Subject: [westpa-users] Using westpa to simulate a reaction with about 35 kcal/mol activation energy

 

--
You received this message because you are subscribed to the Google Groups "westpa-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to westpa-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/westpa-users/8545be3e-93e1-4127-8391-7941767d62e6n%40googlegroups.com.

[Lillian Chong]

Moutoushi,

I agree with Dan that the 35 kcal/mol activation free energy seems unphysical. If you are solely after a PMF (or free energy profile?), you may want to stick with umbrella sampling and apply the WHAM approach to your results. In principle, the WHAM approach would provide a free energy profile. WE, on the other hand, is useful for generating unbiased pathways.

Additional features of your protocol to consider are the appropriate QM level of theory and size of your QM region, which will greatly impact the computational cost of your simulation--not to mention the required 1-fs time step (vs. longer time steps that can be used for fully classical MD simulations).

If you would like any feedback on initial WE efforts, I encourage you to consider the WE data club that Dan mentioned.

All the best,

Lillian

--

You received this message because you are subscribed to the Google Groups "westpa-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to westpa-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/westpa-users/8545be3e-93e1-4127-8391-7941767d62e6n%40googlegroups.com.

--

Lillian T. Chong           
Associate Professor    
Department of Chemistry
University of Pittsburgh
219 Parkman Avenue
Pittsburgh, PA 15260
(412) 624-6026

[moutos...@gmail.com]

Dear Dr. Zuckerman and Dr. Chong,

Thank you very much for your suggestion. I would like to join the data club. I used umbrella sampling with one dimensional projection of reaction coordinates for a proton transfer reaction and I wanted to try to use WESTPA to get reaction kinetics. Like you suggested, I just have to try it first and see if WESTPA can find the path and maybe get the kinetics. I will also closely look at the papers you suggested.

Thank you.

Regards-
Moutoshi

[moutos...@gmail.com]

Dear Dr. Zuckerman and Dr. Chong, 

I am trying to simulate a proton transfer reaction using qmmm. I have the PMF from umbrella simulation and was trying to generate the same PMF with WESTPA. 

I used 1ps simulation time and ran for about 600 iteration steps with 0.1 bin spacing. The flux started increasing after some steps, then started decreasing , then again started increasing (attached 'flux-plot-1ps'). 

I also tried 5ps with 0.5 bin spacing and 500 iterations. The flux plot-5ps showed same behavior (kept increasing and decreasing). 

The PMFs (attached 'PMF-WESTPA') for both 1ps and 5ps runs were not able to catch the 12 kcal/mol drop after reaching the peak (right part of the PMF). 

The unbound state was at -2.9 A and bound state was at +2.9 A. 

What can I do so that WESTPA can catch the 12 kcal/mol drop in PMF? 

I look forward to hearing from you. 

I appreciate your help.

Regards-

Moutoshi

[Daniel Zuckerman]

Moutoshi, glad to see you're getting preliminary data - with successful events if I understand correctly.  I think you should present this in WE data club because email will not work well to understand and advise on a complicated problem.  I will write to you separately to arrange.  --dz

---

From: westpa...@googlegroups.com <westpa...@googlegroups.com> on behalf of moutos...@gmail.com <moutos...@gmail.com>

Sent: Wednesday, January 27, 2021 5:09 PM
To: westpa-users
Subject: [EXTERNAL] Re: [westpa-users] Using westpa to simulate a reaction with about 35 kcal/mol activation energy

 

To view this discussion on the web visit https://groups.google.com/d/msgid/westpa-users/34f8ed2d-871b-4bb6-a149-e5752c063915n%40googlegroups.com.