[gmx-users] Possible to use SETTLE for non-water molecules or switch off LINCS/SHAKE?

[Mark Abraham]

Hi,

On Mon, Mar 12, 2018 at 4:19 PM Douwe Pollmann <douwepollmann at gmail.com>
wrote:

> Dear Gromacs users,
>
> I am currently working on a coarse-grained model where I want to fix atoms
> (Cb's) on a certain distance from Ca-Ca bonds, so that they can rotate
> freely. This freedom will be restricted by a dihedral that determines the
> relative orientation between the Cb's.
> I am using distance constraints or (non-chemical) bonds between a Cb and
> the adjacent Ca's, so that a fixed triangle is formed. I assumed that with
> this method the distance from Cb to the Ca-Ca bond should be correct and
> that my problem was solved. However, sometimes my simulations crash, and it
> appears that both constraint algorithms (LINCS and SHAKE) are not able to
> deal with angle constraints / 3 atoms bonded in a triangle.
>

You should double check the documentation of the mdp options relevant to
these algorithms so that you know you are using them correctly - triangles
of constraints are more demanding than uncoupled ones.

You should also make sure that your exclusions are appropriate. If the Cb
atom is having a non-bonded interaction with the Ca atoms (or their
neighbours) then you make life hard on the constraint implementation.


> I came up with two (maybe very strange) idea to solve this problem:
> 1: There is a third algorithm to solve constraints, called SETTLE, but it
> is stated almost everywhere that it is useful for water molecules. Do you
> know if it is possible to use SETTLE in other molecules than water, for
> example in my system? The Cb would be the "O", and the Ca's the "H's".
>

No, SETTLE is specific to water, ie. molecules with three heavy atoms.


> 2: Is it possible to disable the constraint solver, so that neither Lincs
> and Shake are used? I am asking this because someone mentioned that the
> solvers are meant to reduce the very fast vibrations of light atoms, so
> that a larger mass on my Cb's will also solve the problem.
>

If you don't want constraints, don't ask for any :-) You also won't get
your fixed triangle. You can of course do that, with due regard to your
timestep and validation of your module.

For the method I want to use I really need the Cb's at fixed distances from
> the Ca's and the Ca-Ca bond, but I cannot think of any way without
> triangular constraints that fulfils these requirements.
>
> Does any of you know the answers to the two possible solutions, or a way to
> fix the distance between Cb and the Ca-Ca bond, while maintaining the free
> rotation?
>

The ideal answer would be to use a virtual site in GROMACS, but there's no
way of constructing the Cb atom off the line of the two Ca atoms. There are
ways to construct the Cb atom on the line, or various ways from the
positions of three or more other atoms, but nothing is clearly suitable for
your case, unfortunately.

Mark

Thank you in advance for any help!
>
> Kind regards,
> Douwe Pollmann
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>