Re: Umbrella sampling of a protein through a membrane

From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Fri Apr 12 2019 - 09:12:38 CDT

Hello José, let's assume that the protein always folds/unfolds to reach its
equilibrium structure wherever it's located (water, membrane head groups,
membrane interior, ...): one has to take your word for it.

The issue is how well is the center of mass describing the permeation
pathway: if, for example, the protein unfolds upon binding the lipids'
surface, it's center of mass position will change as an effect of the
unfolding, and make it appear as if the protein was further away from, or
deeper into the membrane. The result will be that a given value of Z there
will characterize more than one of the relevant states. This issue has
nothing to do with choosing the optimal intervals for ABF, which assumes
that you are using a reaction coordinate whose values are able to describe
distinct states.

So there are two suggestions here:

1) Verify that the reaction coordinate is well defined, for example by
extracting frames near an energy barrier, and verifying that multiple
copies of the system will evolve towards both sides of the barrier. You
will only need short unbiased MD runs for this.

2) Derive the windowing strategy with a small rigid molecule instead of the
protein. This will show any issues much more quickly than a protein, which
is going to diffuse orders of magnitude slower and will certainly require
much longer times per window.

Giacomo

On Fri, Apr 12, 2019 at 2:50 AM Villalain Boullon, Jose <jvillalain_at_umh.es>
wrote:

> Dear JC,
>
>
>
> Thank you very much for your response.
>
>
>
> For the SMD simulation the protein, folded, is inserted in the centre of
> the membrane. Then it is applied a relatively small force to the alpha
> carbons in both the z- and z+ directions, together with a small restraint
> to the lipids to maintain the membrane integrity. The protein, passing
> through the membrane, unfolds and then it folds again when in the solvent--0000000000004e5264058655e7c5--

This archive was generated by hypermail 2.1.6 : Sun Sep 15 2019 - 23:20:30 CDT