Re: [AMBER] Estimating protein-protein interactions

From: Daniel Roe <>
Date: Tue, 6 Jul 2021 09:32:42 -0400


On Wed, Jun 30, 2021 at 9:31 AM Matthew Guberman-Pfeffer
<> wrote:
> So, even in the simulation, the VDW term is positive. This puzzles me.

Remember, positive or negative energy terms (or even potential energy)
on their own don't matter so much - the zero point is pretty
arbitrary. What matters is the difference in energy between
conformations, and the free energy of the system (enthalpy *and*
entropy, roughly how the energy tends to be distributed). Just because
the van der Waals term is positive on its own doesn't necessarily mean
a whole lot.

If you're trying to look at the "stability" of something, you are
talking about free energy, not just energy (enthalpy). So you'll need
to do some sort of averaging. You could look at conformational
distributions or some other structural property (e.g. how the average
distance between the proteins or the number of contacts changes with
different conditions) to try to get at the answer indirectly, or you
can try to get at it more directly by averaging snapshot energies
similar to what Niel Henriksen et al. did in this paper:

Or you could try to get at the free energies via something like
umbrella sampling, but defining a reaction coordinate between the
interfaces could be a bit tricky and could be tough to converge.

Anyway, hope this helps,


> Did this help to clarify my question? Please let me know if I should provide more information.
> Best,
> Matthew
> > On Jun 30, 2021, at 9:01 AM, David A Case <> wrote:
> >
> > *Message sent from a system outside of UConn.*
> >
> >
> > On Mon, Jun 28, 2021, Matthew Guberman-Pfeffer wrote:
> >
> >>
> >> I am trying to compute how the stability of a protein-protein interface
> >> changes with protonation of the acidic groups within a 10 Å distance of
> >> the interface. I’ve computed the non-bonded interactions for MD snapshots
> >> using either the linear interaction method with some dielectric constant
> >> specified (4 or 80) or using esander. One aspect that confuses me is that
> >> the VDW energies are positive. I was expecting strongly negative values. Is
> >> there just a sign convention I’m missing?
> >
> > It's not clear what you are computing, or exactly how. How do the numbers
> > you get compare to the values that are printed in the mdout file? Are you
> > only computing a subset of the LJ interactions?
> >
> > ....dac
> >
> >
> > _______________________________________________
> > AMBER mailing list
> >
> >;data=04%7C01%7C%7Cd2b3944ace404d8a2da008d93bc74574%7C17f1a87e2a254eaab9df9d439034b080%7C0%7C0%7C637606549307248214%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=NhAukwOWBi0uUY6L8IWkN9ZHCNpZnSFSuRdYIu26TA4%3D&amp;reserved=0
> _______________________________________________
> AMBER mailing list

AMBER mailing list
Received on Tue Jul 06 2021 - 07:00:02 PDT
Custom Search