# Umbrella sampling position restraining force

**URL:** <https://gromacs.bioexcel.eu/t/umbrella-sampling-position-restraining-force/10125>\
**Category:** User discussions\
**Tags:** umbrella-sampling\
**Created:** [September 8, 2024, 2:36pm UTC](https://gromacs.bioexcel.eu/t/umbrella-sampling-position-restraining-force/10125 "2024-09-08T14:36:20Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![ximi](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/ximi/32/3715_2.png) [@ximi](https://gromacs.bioexcel.eu/u/ximi)\
**Post date:** [September 8, 2024, 2:36pm UTC](https://gromacs.bioexcel.eu/t/umbrella-sampling-position-restraining-force/10125/1 "2024-09-08T14:36:20Z")

</div>

GROMACS version: 2022.2  
GROMACS modification: No  
Dear gromacs forum,

I am using GROMACS to simulate binding between a DNA binding domain and DNA. Following the umbrella sampling [tutorial](https://www.mdtutorials.com/gmx/umbrella), I pulled the binding domain away from DNA, selected configurations (evenly spaced) to perform NPT and md production run. I used the COM distance as a reaction coordinate.

I noticed that the final histogram for calculating WHAM depends on the force constant that is used to restrain molecules. If using a value 1000 KJ/mol/nm^2 (as used in the tutorial), the two molecules will soon collapse to bind, resulting in a very uneven histogram. To avoid this, should I use an arbitrarily large force constant? Also it seems that the force constant can be varying depending on the reaction coordinate. Are there recommendations on parameter selections? I greatly appreciate any suggestion for references.

Thanks in advance.

---

<div class="post-metadata">

**Author:** ![MagnusL](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/magnusl/32/2380_2.png) [@MagnusL](https://gromacs.bioexcel.eu/u/MagnusL)\
**Post date:** [September 9, 2024, 7:24am UTC](https://gromacs.bioexcel.eu/t/umbrella-sampling-position-restraining-force/10125/2 "2024-09-09T07:24:07Z")

</div>

In steep regions of the PMF you might need a higher force constant and/or more windows. There are no clear rules of thumb, but it’s good to check that the histograms are close to bell shaped and overlap properly. For more theory you might want to read [DOI: 10.1021/ct500504g](https://pubs.acs.org/doi/10.1021/ct500504g).

---

<div class="post-metadata">

**Author:** ![ximi](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/ximi/32/3715_2.png) [@ximi](https://gromacs.bioexcel.eu/u/ximi)\
**Post date:** [September 9, 2024, 7:50am UTC](https://gromacs.bioexcel.eu/t/umbrella-sampling-position-restraining-force/10125/3 "2024-09-09T07:50:17Z")

</div>

Thanks a lot for the reply. The linked paper is very helpful!