On Wed, Jul 03, 2019, James Starlight wrote:
>
>Normallly, using PME electrostatic cutoff around 10 angstr with
>Amber99sb, I have found it as a default option with Amber99sb force
>field. However, I have seen also that people se smaller cut-offs
>(around 9) with amber14sb force field, which probably increase
>performance of the calculations.
>What the values of "cut" should be used assuming that I am dealing
>with the modeling of the water soluble protein having stable
>Jelly-roll domain joinded be several long loops. In terms of the
>structure, these flexible segments are enriched with polar/charged
>residues, which may be of the functional importance for the stability/
>activity of the protein?
The value of "cut" has (almost) no effect on electrostatics: it just
determines how much of the electrostatic interaction is computed in
"direct" space vs. "reciprocal" space. The total is very nearly
invariant to the cutoff used.
The importance of "cut" is that it also determines where Lennard-Jones
interactions are ignored, and this can have an effect on structure. The
biggest effects are in lipid bilayers, where the nature of dispersion
interactions is very different in the bilayer plane, compared to the
water region perpendicular to the bilayer. Other very anisotropic
systems might see similar dependencies on "cut".
The field is slowly moving towards adopting a PME-like splitting of LJ
interactions, which would effectively have no cutoff. But for now, the
default value in Amber is 8 Ang., but values of 9 or 10 are often
chosen. I don't know of any definitive study of what happens with
globular proteins, but I hope more informed readers of this list will
chime in with relevant information. (I personally suspect that there is
no general answer, but I've been wrong lots of times in the past.)
...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jul 03 2019 - 07:30:02 PDT
