From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Mon Jan 17 2022 - 12:28:48 CST
Hi Victor,
----- On 14 Jan 22, at 18:44, Victor Hue <victorcolour_at_gmail.com> wrote:
> Dear all,
> I'm using colvars module in LAMMPS, but I believe I will get more help from here
> for enhanced sampling module.
This is a little irregular, but your questions are relevant to the NAMD community as well, so it's fine with me.
> I have 2 questions:
> Question 1
> I have been using umbrella sampling methods for the free energy analysis of
> polymer adsorption from a solid surface. Below are some of the US simulation
> details:
> Distance from the polymer COM to the solids surface: 16A,
> Total window stages: 81 windows
> Window spacing interval: 0.2A between successive windows
> Equilibration time for each window: 6ns (Total all windows 486ns)
> Force constant: 100 kcal/mol.A^2 (after rescale with width 0.1)
> harmonic {
> name h_pot
> colvars dist
> forceConstant 1
> centers {different window stage centre}
> writeTISamples
> outputCenters yes # Write the current centers to the trajectory file
> }
> Using a post trajectory processing tool (WHAM analysis tool by Grossfield), I'm
> able to produce a reasonable PMF profile. However, with different initial
> starting position of the polymer, the PMF profile produced can be quite
> different. Therefore, 10 runs of different initial position are repeated and to
> be averaged for a single polymer case and the results are included in the
> document in the link. The 2 different figures are for the same type of polymer,
> but in neutral and ionized form.
> [
> https://urldefense.com/v3/__https://docs.google.com/document/d/1cahWjFDRIKhHDSAFSS4r-pz1RnP6te15/edit?usp=sharing&ouid=117442070509093491155&rtpof=true&sd=true__;!!DZ3fjg!ucE80VxDdFKE34FKOcV0mXBgLtg0Qxj2pItCmZySYoj_mXjHH4z79pPFWIAwfGC94Q$
> |
> https://urldefense.com/v3/__https://docs.google.com/document/d/1cahWjFDRIKhHDSAFSS4r-pz1RnP6te15/edit?usp=sharing&ouid=117442070509093491155&rtpof=true&sd=true__;!!DZ3fjg!uckMFrN0d1RpmiLqXvi35peFGVil2D-KfbVE07L_QqWbZ5Odz6ZuSbMdM491qFDxMw$
> ]
> It is clear that the neutral polymer has a more consistent PMF profile while the
> ionized polymer can be very inconsistent. I wonder what is the reason causing
> the ionized polymer to have so different range of PMF value? To elaborate more,
> sodium ions are put in the ionized polymer cases to ensure electroneutrality.
> The run 1-5 has smaller PMF value (sodium ions are put near to polymer), while
> run 6-10 seems to have deeper PMF value (sodium ions are put near to solid
> surface).
> I wonder is that because of sodium ions, or any simulation details can I change
> to improve and make it more consistent?
I suspect that the ions are responsible for this. The interactions of a charged polymer with the surface will depend on the position of the counter-ions, and those with fluctuate slowly, introducing noise in your measurements. One option is to increase the ion concentration to improve the sampling. In any case, you should monitor the ion-polymer interactions.
> Question 2
> With the built-in US method in colvars module, ( targetNumSteps and
> targetNumStages ), the PMF profile can be produced from the command writeTIPMF
> . However, is the PMF produced the true PMF, or it contains the biased
> potential? Asking this because I try to change a force constant value from 100
> to 200 kcal/mol.A^2. As a quick analysis (1ns simulation), the PMF profile from
> the command has no difference for a simpler single water molecule, but the PMF
> profile produced is deeper for the polymer with higher force constant.
Could this be a non-equilibrium effect? What happens if you pull slower? (or faster)
Best,
Jérôme
This archive was generated by hypermail 2.1.6 : Tue Dec 13 2022 - 14:32:44 CST