# How can I get smooth PMF from umbrella sampling

**URL:** <https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629>\
**Category:** User discussions\
**Created:** [February 9, 2022, 4:34am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629 "2022-02-09T04:34:49Z")\
**Posts on this page:** 13\
**Page:** 1

<div class="post-metadata">

**Author:** ![mijiddorj](https://avatars.discourse-cdn.com/v4/letter/m/779978/32.png) [@mijiddorj](https://gromacs.bioexcel.eu/u/mijiddorj)\
**Post date:** [February 9, 2022, 4:34am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/1 "2022-02-09T04:34:49Z")

</div>

GROMACS version: 2018  
GROMACS modification: No

Dear GMX users,

I performed two umbrella sampling simulations of interaction of two short peptides for 20 ns each window. I got some weird PMF figures as I attached. It means that there are sharp curves for both simulations. How can I get smooth PMF?

Second, the converged values of these PMF is around -3 kcal/mol. How reliable are these curves?

Best regards,  
Batsaikhan

 ![profile](https://europe1.discourse-cdn.com/flex017/uploads/bioexcel1/original/2X/c/c4f11bd86a45d20461dda34770797eaf604c11ff.jpeg)

---

<div class="post-metadata">

**Author:** ![jalemkul](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/jalemkul/32/18_2.png) [@jalemkul](https://gromacs.bioexcel.eu/u/jalemkul)\
**Post date:** [February 9, 2022, 4:45am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/2 "2022-02-09T04:45:43Z")

</div>

Look at the histograms. You probably have a few regions of poor sampling/poor overlap.

---

<div class="post-metadata">

**Author:** ![mijiddorj](https://avatars.discourse-cdn.com/v4/letter/m/779978/32.png) [@mijiddorj](https://gromacs.bioexcel.eu/u/mijiddorj)\
**Post date:** [February 9, 2022, 4:51am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/3 "2022-02-09T04:51:57Z")

</div>

Dear prof. Justin,  
Thank you very much for your prompt reply. I attached one of the histograms. It looks much overlapped. Should I increase simulation time of the windows.

Best regards,  
Batsaikhan

 ![histo](https://europe1.discourse-cdn.com/flex017/uploads/bioexcel1/original/2X/8/856435e555d875bebe8c38f5eef7bdff35c90796.jpeg)

---

<div class="post-metadata">

**Author:** ![jalemkul](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/jalemkul/32/18_2.png) [@jalemkul](https://gromacs.bioexcel.eu/u/jalemkul)\
**Post date:** [February 10, 2022, 2:25am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/4 "2022-02-10T02:25:00Z")

</div>

You have way too many windows that are almost redundant. There is really uneven spacing across the reaction coordinate. You need to refine your window spacing, force constants, or both. And you may still need more sampling time after you do that.

---

<div class="post-metadata">

**Author:** ![mijiddorj](https://avatars.discourse-cdn.com/v4/letter/m/779978/32.png) [@mijiddorj](https://gromacs.bioexcel.eu/u/mijiddorj)\
**Post date:** [February 12, 2022, 2:59am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/5 "2022-02-12T02:59:20Z")

</div>

Dear prof. Justin,

Thank you very much for your clear explanation.

Best regards,  
Batsaikhan

---

<div class="post-metadata">

**Author:** ![mijiddorj](https://avatars.discourse-cdn.com/v4/letter/m/779978/32.png) [@mijiddorj](https://gromacs.bioexcel.eu/u/mijiddorj)\
**Post date:** [March 23, 2022, 1:07am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/6 "2022-03-23T01:07:27Z")

</div>

Dear prof. Justin,

Hello, thank you for your time.  
I performed umbrella sampling simulations for dimer structure of a short peptide.  
When I perform WHAM analysis, the histogram is good enough as following. However, the PMF plot is weird. How can I get good PMF curve? Please advice me.

 ![histo](https://europe1.discourse-cdn.com/flex017/uploads/bioexcel1/original/2X/4/4f43816a30758c2ce2846a04c6475841965f0e45.jpeg)  
 ![profile](https://europe1.discourse-cdn.com/flex017/uploads/bioexcel1/original/2X/e/ed5407ea2b887ac15a295c7a264bf43bba0176fd.jpeg)

Best regards,  
Batsaikhan

---

<div class="post-metadata">

**Author:** ![jalemkul](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/jalemkul/32/18_2.png) [@jalemkul](https://gromacs.bioexcel.eu/u/jalemkul)\
**Post date:** [March 25, 2022, 6:32pm UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/7 "2022-03-25T18:32:30Z")

</div>

Looks fine to me. What do you think is the problem?

---

<div class="post-metadata">

**Author:** ![mijiddorj](https://avatars.discourse-cdn.com/v4/letter/m/779978/32.png) [@mijiddorj](https://gromacs.bioexcel.eu/u/mijiddorj)\
**Post date:** [March 27, 2022, 8:13am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/8 "2022-03-27T08:13:17Z")

</div>

Dear prof. Justin,

Thank you very much for your reply. I have tried many times to get this kind of histograms using different forces and different simulation times. Some of calculations showed the higher converging values in profile plot around 1 or 2 kcal/mol. However I used very strong pulling force, I observed uneven distributions in strong pulling simulations.

Therefore, I expected that the profile curve should be converged around positive E values.

Thank you very much again.

Best regards,

Batsaikhan

---

<div class="post-metadata">

**Author:** ![abdelmoujoud.faris](https://avatars.discourse-cdn.com/v4/letter/a/bbe5ce/32.png) [@abdelmoujoud.faris](https://gromacs.bioexcel.eu/u/abdelmoujoud.faris)\
**Post date:** [October 26, 2023, 6:03am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/9 "2023-10-26T06:03:12Z")

</div>

pls how can i run PMF at gromacs

---

<div class="post-metadata">

**Author:** ![abdelmoujoud.faris](https://avatars.discourse-cdn.com/v4/letter/a/bbe5ce/32.png) [@abdelmoujoud.faris](https://gromacs.bioexcel.eu/u/abdelmoujoud.faris)\
**Post date:** [October 26, 2023, 6:04am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/10 "2023-10-26T06:04:25Z")

</div>

Please can I have the PMF script to run it on gromacs and thank you.

---

<div class="post-metadata">

**Author:** ![abdelmoujoud.faris](https://avatars.discourse-cdn.com/v4/letter/a/bbe5ce/32.png) [@abdelmoujoud.faris](https://gromacs.bioexcel.eu/u/abdelmoujoud.faris)\
**Post date:** [October 26, 2023, 8:09am UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/11 "2023-10-26T08:09:25Z")

</div>

Hello dear professor. I have performed an MD simulation for 200 ns, and now I’m wondering if there is a script that I can use for HPC to calculate the potential of mean force (PMF) using umbrella sampling. Thank you.

---

<div class="post-metadata">

**Author:** ![Eduardo1](https://avatars.discourse-cdn.com/v4/letter/e/838e76/32.png) [@Eduardo1](https://gromacs.bioexcel.eu/u/Eduardo1)\
**Post date:** [June 5, 2024, 5:41pm UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/12 "2024-06-05T17:41:37Z")

</div>

Dear jalemkul,

How do I need to enhance sampling in regions with poor sampling along the reaction coordinate? In my work, I run 180 simulations of 100 ns (umbrella sampling) to represent the translocation process of a polymer through a membrane. The umbrella histograms shows a poor sampling around 4 nm:

 ![Poor-sampling-2](https://europe1.discourse-cdn.com/flex017/uploads/bioexcel1/original/2X/7/76d2892002f6d4a3b6075b4059b9814bd14f9dfe.png)

I tried to run more simulations, but they did not sample the region around 4 nm.

---

<div class="post-metadata">

**Author:** ![jalemkul](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/jalemkul/32/18_2.png) [@jalemkul](https://gromacs.bioexcel.eu/u/jalemkul)\
**Post date:** [June 5, 2024, 5:56pm UTC](https://gromacs.bioexcel.eu/t/how-can-i-get-smooth-pmf-from-umbrella-sampling/3629/13 "2024-06-05T17:56:39Z")

</div>

You need more windows to fill the gap, perhaps with a stronger force constant if it is a very high-energy region of the reaction coordinate.
