normal Membrane systems with PS lipid heads: W or PW?

  • Prunotto
  • Prunotto's Avatar Topic Author
  • Offline
  • Fresh Boarder
4 years 1 month ago #4518 by Prunotto

I am running a membrane simulation using different kinds of lipids. When I solvated the system with polarizable water (setting dielectric constant at 2.5) and ran the minimization, I got several LINCS warnings. Despite of this, the minimization was able to finish with a final max force of roughly 10^3.
Nevertheless, the following equilibration crashed (because of the LINCS warnings).

On the contrary, when I solvate the system with ordinary MARTINI water (setting dielectric constant at 15), both the minimization and the equilibration end well.

The system is composed by several lipids, including a concentration of negatively charged lipids (PS). Do you think this might be the problem?

If I use ordinary MARTINI water instead of PW, do you think that I will lose some information?

Many thanks,


Please Log in or Create an account to join the conversation.

4 years 1 month ago - 4 years 1 month ago #4523 by mnmelo
Hi Alessio,

It's not uncommon that systems with many constraints fail to minimize properly, and that might be the case of your polarizable waters. Their dummy charges are bound to the water's LJ center by constraints.

The easiest way around this is to edit your particle definition file (martini_v2.2P.itp). At the very end you'll find the parameters for polarizable water (moleculename PW). You'll have to comment out the directive "[ constraints ]" and the lines under it, and uncomment the directive "[ bonds ]" and the lines under it. When you're done your PW parameters should look like this:

[ moleculetype ]
; molname       nrexcl
PW              1

[ atoms ]
;id type resnr residu atom cgnr   charge
 1   POL   1    PW     W    1      0
 2   D     1    PW     WP   1      0.46
 3   D     1    PW     WM   1     -0.46

;  i     j   funct   length
;   1     2    1       0.14
;   1     3    1       0.14

; for minimization purposes constraints might be replaced by stiff bonds:
;  i     j   funct   length   force const.
   1     2    1       0.14    50000
   1     3    1       0.14    50000

;   i    j   k   funct  angle    fc
    2    1   3    2     0.0     4.2

   1     2  3
   2     3

Minimization should now run much better. Don't forget to change the definitions back for your equilibration and production runs! (We'll probably provide an .itp file in the future where this switchng can be done more easily using grompp's #defines. Feel free to implement it yourself, if you're comfortable with it.)

On your second question:
Using regular Martini water will give you less accurate electrostatic interactions, but at an increased performance. How relevant this is is system dependent and it's up to you to decide (and later show that you made the adequate choice).

As a rule of thumb, if you expect solvation of certain groups to play an essential role in the behavior of your system, then do use PW; otherwise it may be best to start with regular water and only move to PW if you identify shortcomings you suspect are related to electrostatic interactions.

As an example: regular Martini water can be used with most bilayer systems without ill-effects regarding membrane properties. However, when highly-charged compounds are simulated in interaction with these bilayers, then it is best to use PW.

Good luck,
Last edit: 4 years 1 month ago by mnmelo. Reason: Code formatting

Please Log in or Create an account to join the conversation.

Time to create page: 0.089 seconds