Ooohh.. You may have found a bug (negative feature?) in the way the montecarlo barostat is implemented. We rarely use semi-isotropic now which means it was probably not tested against the monte carlo barostat. I think I know where the problem might be. Let me take a look and we'll see if we can put in a quick fix - although it might need testing to make sure it works correctly.
The problem you are seeing with fully anisotropic is that because the X,Y and Z directions are not coupled there is nothing preventing the box from shrinking completely in one dimension. I am thinking that a way to improve this might be to have a very weak coupling between the X and Y dimensions but I am not sure if that would be formally correct or not.
All the best
Ross
> On Jul 1, 2015, at 3:01 PM, Naomi Rose Latorraca <nlatorra.stanford.edu> wrote:
>
> Hi Amber users,
>
> I've been using Amber software to run simulations of a small membrane protein embedded in a POPC bilayer. The system, which is parameterized with charmm parameters, is about 35,000 atoms in size and has dimensions of 70 x 70 x 73 angstroms.
>
> For pressure control settings, I've been using anisotropic coupling (npt=2) with the Monte Carlo barostat (barostat=2). However, on the time scale of tens to hundreds of nanoseconds, we observe lengthening of the system in a single lateral dimension. Upon visualizing the periodic box images in VMD, we see that boxes are shearing against each other in the lateral dimensions in the bilayer and that protein images are getting quite close to the edges of each individual box.
>
> Due to the concern that the protein may see itself in adjacent images, we switched to using semi-isotropic control, with the options npt=3, csurften=3, and gamma_ten=0.0. However, even with these settings changed, I still observe lengthening of the system in a single lateral dimension, which I wouldn't expect with semi-isotropic pressure coupling. I then switched my barostat to Berendsen (barostat=1) and found that I observe only slight expansions and contractions in volume during simulation, and the x-y dimensions are coupled during these changes, as I would expect to see with true semi-isotropic coupling. How is semi-isotropic coupling implemented in Amber14, and is it necessary to use the Berendsen barostat in conjunction with this setting? Furthermore, what conditions merit the use of anisotropic vs semi-isotropic coupling?
>
> Best,
> Naomi Latorraca
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jul 01 2015 - 16:00:02 PDT
