Re: [AMBER] Few queries about membrane protein simulations

From: Jason Swails <>
Date: Mon, 8 Jun 2015 12:04:11 -0400

On Mon, Jun 8, 2015 at 10:45 AM, anu chandra <> wrote:

> Dear Amber users,
> Though I have been using Amber for protein simulations, I am here trying to
> use Amber for my membrane protein simulations. Since I am new to membrane
> protein simulations, I would like to get your suggestions for few of my
> quires, which will helpful for me to set the simulation on.
> My queries are,
> 1. I wish to use lipid 14 and FF14SB force fields for membrane and protein
> respectively. Will it be okay? Do I have to use any correction terms here?
> I will be using POPC membrane.

​No, the force fields should be compatible. (I'm not sure what you mean by
"correction term" -- in general, force fields are either compatible or not;
there is no 'correction term' to make incompatible force fields compatible
that I know of).

> 2. I have noticed in literature that , for microsecond long simulations,
> sometimes a 3 fs or 4 fs time steps were used. How safe to use such a large
> simulation time step, though it reduce the use of computations resource?

​3 or 4 fs time steps are used with hydrogen mass repartitioning. From
everything I've been able to tell, as long as you use HMR with a 4 fs
timestep, the results are statistically indistinguishable from standard
masses with a 2 fs time step.

> 3. The membrane protein, I am working with, transport ions depend on the
> concentration gradient. In order to implement this in my simulation, I am
> planning to add ions ( for e.g KCl) in one of the leaflet to see the
> movement of ions to the other leaflet through the pore. Am I making sense
> here, by restricting ion distribution to only one of the leaflet rather
> than randomly placing in the entire simulation box (usually do for protein
> simulation) at the starting of simulation?

​Possibly. It sounds reasonable to me. Just take care of the implications
of periodicity -- for instance, if you have only a single bilayer with ions
on one side of the bilayer, it can simply move "up" to the unit cell above
without crossing through any ion channels in the membrane itself.

> 4. How can I build slab geometry boundary condition in Amber to maintain
> ​​
> ion asymmetry? (

I think Amber supports only full 3-D periodic boundary conditions or no
boundary conditions.

> ​
> 5. If I would like to do membrane protein simulations with an applied
> ​​
> electric filed along Z-direction, how can I implement the electric field in
> Amber simulations?

​Also not implemented, to my knowledge.

> 6. Is it possible to do PMF calculation with classical MD trajectories or
> do I want to carried out umbrella sampling separately for PMF?

​You can always compute an unbiased PMF using raw probabilities from MD
simulations. You only need to resort to something like umbrella sampling
if parts of the reaction coordinate have poor or no sampling.

> 7. Though I wish to use Amber for membrane protein simulation, some one
> kindly provide some reference for membrane protein simulation where Amber
> ​​
> used? ( sorry, I couldn't get one)

​Look at the reference papers for the lipid force fields.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Mon Jun 08 2015 - 09:30:02 PDT
Custom Search