On Fri, May 29, 2020, Timothy Schutt wrote:
>
>Is it an okay approximation to use igb=1 and extdiel=*dielectric value* to
>represent a non-aqueous solvent implicitly?
"okay" is a tough thing to answer here. Some points to consider:
1. Non-aqueous solvents are charactized by more than just a dieletric
constant; with gb, changing extdiel changes way in which Coulomb
interactions are screened. This does not capture all of the differences
between solvent.
2. Amber's GB models have not been tested (much) with other solvents.
Maybe I'm unaware of some of this work, and I hope others will chime in
with appropriate references. So, as far as I know, one doesn't know if
your proposed scheme will yield "reliable trends" for different
solvents. On top of this, you are proposing a kind of hydbrid model,
with one explicit solvent molecule and the rest being implicit; this is also
likely to require careful testing.
That said, this is research, after all. It might well work, and has
some reasonable physics in it, especially if your solvent molecules are
fairly small and rigid. n-octanol as a solvent would be stretch, since
it is floppy and has both polar and non-polar parts. It's a balance
between how much time your have and how accurate your results need to
be.
An alternative would be to stick with explicit solvation, and use some
enhanced sampling method (replica exchange, self-guilded Langevin,
perhaps one of the GaMD variants) to address the viscosity problem.
....good luck....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Jun 01 2020 - 07:30:03 PDT
